Stress Testing - IMF eLibrary - International Monetary Fund

Stress TestingPrinciples, Concepts, and Frameworks

Editors

Li Lian Ong and Andreas A. Jobst

I n t E r n A t I O n A L M O n E t A r y F u n d

©International Monetary Fund. Not for Redistribution

©2020 International Monetary Fund

Cover design: IMF CSF Creative Solutions Division

Cata loging- in- Publication DataIMF Library

Names: Ong, Li Lian, editor. | Jobst, Andreas A., editor. | International Monetary Fund, publisher.Title: Stress testing: principles, concepts, and frameworks / editors, Li Lian Ong and Andreas A. Jobst.Description: Washington, DC : International Monetary Fund, 2020. | Includes bibliographical references.Identifiers: ISBN 978-1-48431-071-7 (hardback) 978-1-51351-679-0 (pdf) 978-1-51351-688-2 (ePub)Subjects: LCSH: Financial crises. | Banks and banking, International.Classification: LCC HB3725.G84 2020

Disclaimer: The views expressed in this book are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. The boundaries, colors, denominations, and any other information shown on the maps do not imply, on the part of the IMF, any judgment on the legal status of any territory or any endorsement or acceptance of such boundaries.

Please send orders to:International Monetary Fund, Publication Ser vicesP.O. Box 92780, Washington, DC 20090, U.S.A.Tel.: (202) 623- 7430 Fax: (202) 623- 7201

E-mail: [email protected] www.bookstore.imf.orgwww.elibrary.imf.org


http://www.bookstore.imf.org

http://www.elibrary.imf.org

Contents

Foreword ....................................................................................................................................................................................... vtObIAs AdrIAn

Preface .......................................................................................................................................................................................... vii

Abbreviations ............................................................................................................................................................................. ix

Contributing Authors .............................................................................................................................................................. xv

1. stress testing at the International Monetary Fund: Principles, Concepts, and Frameworks ..................................................................................................................... 1AndrEAs A. JObst • LI LIAn Ong

PART I PrinCiPles

2. Macro- Financial stress testing: Principles and Practices ............................................................11HIrOkO OurA • LILIAnA sCHuMACHEr

wItH JOrgE A. CHAn- LAu • DIMItrI g. dEMEkAs • DALE F. grAy • HEIkO HEssE • AndrEAs A. JObst • EMAnuEL

kOPP • SònIA MuñOz • LI LIAn Ong • CHrIstInE sAMPIC • CHrIstIAn sCHMIEdEr • ROdOLFO wEHrHAHn

3. stress tests As a systemic risk Assessment tool ............................................................................55dIMItrI g. dEMEkAs

PART II ConCePts

4. the global Macro- Financial Model: A stress test scenario simulation tool ........................63FrAnCIs VItEk

5. g20 data gaps Initiative II: Meeting the Policy Challenge .........................................................81rObErt HEAtH • EVrIM bEsE gOksu

6. ring- Fencing and Consolidated banks’ stress tests................................................................... 101EugEnIO CEruttI • CHrIstIAn sCHMIEdEr

7. rules of thumb for bank solvency stress testing ....................................................................... 119dAnIEL C. HArdy • CHrIstIAn sCHMIEdEr

8. bank solvency and Funding Cost: new data and new results ............................................. 155stEFAn w. sCHMItz • MICHAEL sIgMund • LAurA VALdErrAMA

9. sovereign risk in Macroprudential solvency stress testing ................................................... 183AndrEAs A. JObst • HIrOkO OurA


Underline

Underline

Contentsiv

10. revisiting risk- weighted Assets: why do rwAs differ across Countries and what Can be done about It? ................................................................................. 229VAnEssA LE LEsLé • AndrEAs A. JObst

11. A new Heuristic Measure of Fragility and tail risks: Application to stress testing .............................................................................................................. 261nAssIM nicholas tALEb • ELIE CAnEttI • TIdIAnE kIndA • ELEnA LOukOIAnOVA • CHrIstIAn sCHMIEdEr

12. How to Capture Macro- Financial spillover Effects in stress tests ......................................... 277HEIkO HEssE • FErHAn sALMAn • CHrIstIAn sCHMIEdEr

13. real and Financial Vulnerabilities from Cross- border banking Linkages ........................... 299kyungHun kIM • SrObOnA MItrA

14. Credibility and Crisis stress testing ................................................................................................. 313LI LIAn Ong • CEyLA PAzArbAsIOgLu

PART III Frameworks

15. Macroprudential bank solvency stress testing in FsAPs for systemically Important Financial systems .................................................................................... 365AndrEAs A. JObst • LI LIAn Ong • CHrIstIAn sCHMIEdEr

16. Macroprudential bank Liquidity stress testing in FsAPs for systemically Important Financial systems .................................................................................... 403AndrEAs A. JObst • LI LIAn Ong • CHrIstIAn sCHMIEdEr

17. Macroprudential solvency stress testing of the Insurance sector ...................................... 445AndrEAs A. JObst • NObuyAsu sugIMOtO • TIMO brOszEIt

Index .......................................................................................................................................................................................... 515

tOOLkIt COntEnts

the files listed below are available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress -test2-toolkit.

Chapter 9• IMF FsAP sovereign risk stress testing tool (with data input file “data Input”)

Chapter 15• stress testing Matrix (steM)—solvency stress testing Approaches in IMF FsAPs• Presentation templates for IMF FsAP solvency stress test results

Chapter 16• stress testing Matrix (steM)—Liquidity stress testing Approaches in IMF FsAPs• IMF FsAP Liquidity stress testing tool

Chapter 17• stylized summary of Insurance stress testing Approaches in IMF FsAPs and national

supervisory Frameworks• stress testing Matrix (steM)—Insurance stress testing Approaches in IMF FsAPs• stress testing Matrix (steM)—Insurance stress testing Approaches in national

supervisory Frameworks• Presentation template for IMF FsAP Insurance stress test results


https://www.elibrary.imf.org/page/stress-test2-toolkit


Underline

Foreword

What do you do when you need to measure how shocks affect the financial system? You call in stress testers!Stress testing is a crucial tool for measuring financial system resilience. It has been an indispensable component of financial

stability assessments that the IMF staff carries out under the Financial Sector Assessment Program (FSAP). Stress tests have also become an integral part of financial stability analyses done by central banks, supervisory agencies, and many others.

The scope and design of stress tests have significantly evolved since the global financial crisis. The crisis demonstrated that the usefulness and effectiveness of stress testing depend on consistency, comparability, and coherence of methods and models. Stress tests can go wrong if they are based on weak, flawed, or incomplete data, omit crucial risks, or exclude important non-bank financial institutions and financial market infrastructures. At the same time, stress tests can provide extremely powerful insights into the financial system’s vulnerabilities if they are done credibly and transparently. It is therefore important for the design, implementation, and findings of stress tests to be well explained and well understood. That is why a book such as this one— focused on principles, concepts, and frameworks— is incredibly helpful.

The IMF has had a pioneering role in the stress testing of financial systems. Since the introduction of the FSAP by the IMF and the World Bank 20 years ago, the IMF staff has completed more than 350 financial stability assessments, including stress tests, in countries accounting for more than 99 percent of global financial system assets. As a result, the IMF staff has amassed a wealth of hands- on experience with stress testing techniques and their practical applications. Member countries have fre-quently requested IMF technical assistance developing their own stress testing approaches.

This anthology is a follow up to the 2014 book A Guide to IMF Stress Testing: Methods and Models. It assembles additional papers written by individual IMF staff members and their coauthors on the principles and concepts that have shaped the imple-mentation of FSAP stress tests over the past eight years. (The papers collected in the volume reflect the work of the individual authors and do not necessarily represent the views of the IMF, its Executive Board, or IMF management.) They capture much of the evolving anatomy of IMF stress testing by providing insights into the design, risk coverage, and practical implementation of stress tests for financial stability analysis and crisis management. These approaches have, at one time or another, been applied in IMF surveillance of, or technical assistance to, member countries. As in the 2014 book, some of the chapters are accompa-nied by available tools.

While this volume covers a substantial body of work, it is not exhaustive and has not been fully updated. The book does not include all of the studies produced by the IMF staff 1 as well as work in progress and recent developments in stress testing as result of ongoing financial stability assessments. Moreover, empirical analyses underlying the chapters have not been updated, so the chapters should be seen as a series of snapshots capturing various vintages of work by individual IMF staff members and their coauthors, reflecting the issuance dates of the respective papers. All chapters were written before the COVID-19 crisis, and as such do not reflect lessons from the pandemic.

With these limitations in mind, let me highlight some areas of recent emphasis.2 The IMF staff has significantly stepped up work to strengthen the analytical underpinnings of stress tests. Key areas of recent focus include (1) incorporating feedback effects between the financial sector and the real economy; (2) further extending the coverage to nonbank financial institutions and finan-cial markets; (3) improving models of spillovers among institutions and across borders; (4) developing new models for the interac-tion between liquidity and solvency risks; and (5) advancing agent- based modeling. The 2020 Review of the FSAP will be an opportunity to examine and report on some of these improvements as well as on emerging lessons from the COVID-19 crisis.

In addition to these efforts aimed at strengthening the analytical underpinnings of existing stress tests, the IMF staff has also explored new topics. This includes stress tests for emerging risks such as technological disruptions and climate change. Given the evolving nature of this work, these risks have not been covered in this book and also remain a work in progress at the IMF.

Technology- enabled innovation is disrupting the provision of financial services. The evolution of “Fintech” in bank and nonbank financial intermediation could weigh on profitability and liquidity and stoke risk taking. Recent IMF stress tests have started exploring these issues, for example, by examining how the rise of new financial technologies could squeeze the profits of

1 To learn more about IMF stress testing, visit www.imf.org and www.elibrary.imf.org and type “stress testing” or “risk analysis” in the search bar. 2 These are explained more fully in the recent Departmental Paper No. 20/04 titled “Stress Testing at the IMF” at https://www.imf.org/en/Publications

/Departmental-Papers-Policy-Papers/Issues/2020/01/31/Stress-Testing-at-the-IMF-48825.


http://www.imf.org

http://www.elibrary.imf.org

https://www.imf.org/en/Publications/Departmental-Papers-Policy-Papers/Issues/2020/01/31/Stress-Testing-at-the-IMF-48825

https://www.imf.org/en/Publications/Departmental-Papers-Policy-Papers/Issues/2020/01/31/Stress-Testing-at-the-IMF-48825

vi Foreword

existing financial service firms. Separately, the increased digitization of financial services can also expose them to cyber risk with possibly systemic consequences for financial intermediation. For instance, a major cyber event could lead to runs on deposits or claims against insurers. The IMF staff has been working collaboratively with external experts on stress testing for cyber risk.

Climate risk is a major systemic issue on which the IMF staff has been working intensely. FSAP stress tests often capture “physical risks,” such as insurance losses and nonperforming loans associated with storms, floods, and droughts. Over time, these tests have devolved from narrow exercises focusing on nonlife insurance companies to integrated assessments capturing the macro- financial effects. Several ongoing assessments are expanding on this analysis, examining the effects of increased frequency and costs of natural disasters as well as the risks involved in the transition to low- carbon economies. More work is still needed to better capture second- round effects and to fill the major data gaps in this area. The IMF staff collaborates with central banks and others to enhance the analysis of macro- financial transmission of climate risks, including through stress tests.

In sum, stress testing is an exciting, evolving, and important part of the broader effort to ensure financial stability. I trust that this volume will provide a valuable resource for central bankers, financial sector supervisors, other country officials, and academics as well as anyone interested in better understanding the concepts and principles that have been shaping stress testing approaches developed by the IMF staff.

TOBIAS ADRIANFinancial Counsellor and DirectorMonetary and Capital Markets DepartmentInternational Monetary Fund


Preface

This book was produced prior to the COVID-19 crisis. Therefore, it does not reflect the effect of these developments and related policy measures on the IMF staff’s analysis of potential financial stability implications, as well as relevant analytical tools and stress testing approaches. Readers may consult the IMF’s COVID-19 website,1 which includes a tracker of key policy measures2 and staff recommendations with regard to the COVID-19 global outbreak, including on financial sector regulation and supervision,3 as well as a current analysis of financial stability implications in the April 2020 issue of the Global Financial Stability Report.4

However, some aspects of stress testing remain relevant even during the current crisis. Evaluating the extent to which the financial system is resilient to the adverse impact of the crisis underscores the importance of a robust stress testing framework as an integral part of overall risk governance and market surveillance. Stress tests are typically prospective—they help financial institutions and supervisors assess solvency and liquidity risks under severe but plausible scenarios while there may still be time to remedy identified vulnerabilities. However, when shocks have already occurred, such as in the case of the current COVID-19 crisis, the role of stress tests changes. Findings from past stress test exercises on preexisting vulnerabilities (if available) can help inform timely policy decisions as developments unfold—ideally in the form of scenario analyses (potentially in combination with selective data updates), especially if the deterioration in macroeconomic conditions and the impact on the financial system remain highly uncertain. For instance, in the case of the COVID-19 crisis, interventions, such as regulatory forbearance for banks as well as debt moratoria or state guarantees for household and corporate borrowers, cushion the immediate impact of potential impairments but could just postpone potentially serious disruptions to the lending channel. Thus, certain aspects of stress testing covered throughout this book can provide valuable guidance on how to identify downside risks and chart a course of action for safeguarding financial stability during the recovery.

We are grateful to the many contributing authors of this book for their support throughout the process. The papers that make up the chapters have benefited from comments from other IMF staff, academics, market participants, and policymakers, as well as journal editors and referees. We would like to thank the management of the Monetary and Capital Markets Depart-ment, notably, Tobias Adrian, James Morsink, Ratna Sahay, and Martin Čihák, for backing this book. This book has also benefited greatly from the expertise and advice of colleagues in the Communications Department. Last but not least, we would like to thank our indefatigable copy editor, Lorraine Coffey, for her heroic efforts and immeasurable patience in shepherding us and the manuscript to its completion.

1 https://www.imf.org/covid192 https://www.imf.org/en/Topics/imf-and-covid19/Policy-Responses-to-COVID-193 https://www.imf.org/en/Publications/Miscellaneous-Publication-Other/Issues/2020/05/20/COVID-19-The-Regulatory

-and-Supervisory-Implications-for-the-Banking-Sector-494524 https://www.imf.org/en/Publications/GFSR/Issues/2020/04/14/global-financial-stability-report-april-2020


https://www.imf.org/covid19

https://www.imf.org/en/Topics/imf-and-covid19/Policy-Responses-to-COVID-19

https://www.imf.org/en/Publications/Miscellaneous-Publication-Other/Issues/2020/05/20/COVID-19-The-Regulatory-and-Supervisory-Implications-for-the-Banking-Sector-49452

https://www.imf.org/en/Publications/GFSR/Issues/2020/04/14/global-financial-stability-report-april-2020

https://www.imf.org/en/Publications/Miscellaneous-Publication-Other/Issues/2020/05/20/COVID-19-The-Regulatory-and-Supervisory-Implications-for-the-Banking-Sector-49452

This page intentionally left blank


abbreviations

2SLS Two- stage least squares3SLS Three- stage least squaresABS Asset- backed securityAE Advanced economyAfS Available- for- saleAIRB Advanced internal- ratings- basedALM Asset- liability managementAM Advanced marketAP Asia- PacificAQR Asset quality reviewAT AustriaAU AustraliaBBVA Banco Bilbao Vizcaya Argentaria, S.A.BCBS Basel Committee for Banking SupervisionBCP Banco Comercial Português, SABdE Banco de EspañaBE BelgiumBIS Bank for International SettlementsBMA Bermuda Monetary AuthorityBoE Bank of England BoF- PSS2 Bank of Finland payment system simulatorbps Basis pointsBSCR Bermuda Solvency Capital RequirementBU Bottom- upCA CanadaCAR Capital adequacy ratioCBI Central Bank of IrelandCCA Contingent claims analysisCCAR Comprehensive Capital Analysis and ReviewCCP Central counterpartyCDIS Coordinated Direct Investment SurveyCDMs Concentration and distribution measuresCDS Credit default swapCEBS Committee of European Banking SupervisorsCEIOPS Committee of European Insurance and Occupational Pensions SupervisorsCET1 Common equity tier 1CF Cash flowCGFS Committee on the Global Financial System


x Abbreviations

CH SwitzerlandCHF Swiss francCMBS Commercial mortgage- backed securityCoVaR Conditional value at riskCPI Consumer price indexCPIS Coordinated Portfolio Investment SurveyCPPI Commercial Property Price IndicesCPSS Committee on Payment and Settlement SystemsCRA Credit rating agencyCRD Capital Requirements DirectiveCRE Commercial real estateCRR Capital Requirements RegulationCSD Central securities depositoryCT1 Core Tier 1CTE Conditional tail expectationDCC Dynamic conditional correlationDCF Discounted cash flowDCLG Department of Communities and Local GovernmentDD Double dipDE GermanyDEU Deutsche markDFA Dodd- Frank Wall Street Reform and Consumer Protection ActDFAST Dodd- Frank Annual Stress TestsDGI G20 Data Gaps InitiativeDKK Danish kronaDSGE Dynamic stochastic general equilibriumDVI Data integrity and verificationEBA European Banking AuthorityEBRD European Bank for Reconstruction and DevelopmentEC European CommissionECB European Central BankECR Enhanced capital requirementEDF Expected default frequencyEFS Enhanced Fujita scale rate of 5 (>200 mph)EFSF European Financial Stability FacilityEIOPA European Insurance and Occupational Pensions AuthorityEM Emerging marketEMDEs Emerging markets and developing economiesEME Emerging market economyES SpainESM European Stability MechanismESRB European Systemic Risk BoardEU European UnionEUR EuroEVT Extreme value theoryFDIC Federal Deposit Insurance Corporation


Abbreviations xi

FI FinlandFIRB Foundation internal- ratings- basedFMCBG Finance Ministers and Central Bank GovernorsFMI Financial market infrastructureFR FranceFRTB Fundamental Review of the Trading BookFSA Financial Services AuthorityFSAP Financial Sector Assessment ProgramFSB Financial Stability BoardFSIs Financial Soundness IndicatorsFSSA Financial System Stability AssessmentFTSE Financial Times Stock ExchangeFV Fair valueFVCDS Fair value credit default swapFVOAS Fair value option adjusted spreadFW ForwardFX Foreign exchangeGAAP Generally Accepted Accounting PrinciplesGARCH Generalized autoregressive conditional heteroskedasticityGB United KingdomGBP UK pound sterlingGDDS General Data Dissemination SystemGEV Generalized extreme valueGFC Global financial crisisGFF Global Flow of FundsGFM Global macro- financial modelGFS Government Finance StatisticsGFSM Government Finance Statistics ManualGFSR Global Financial Stability ReportGIIPS Greece, Ireland, Italy, Portugal, SpainGIP Greece, Ireland, PortugalGNP Gross national productGR GreeceGSE Government- sponsored entity G- SIB Global systemically important bank G- SIFI Global systemically important financial institutionsH HeuristicHfT Held- for- tradingHICP Harmonized Index of Consumer PricesHtM Held- to- maturityIAA International Actuarial AssociationIAG Inter-Agency Group on Economic and Financial StatisticsIAIG Internationally Active Insurance GroupIAIS International Association of Insurance SupervisorsIB Investment bankIBS International Banking Statistics


Abbreviationsxii

ICP Insurance Core PrincipleIE IrelandIEO Independent Evaluation OfficeIFRS International Financial Reporting StandardsIG Investment grade- relatedIIP International investment positionIN IndiaIOSCO International Organization of Securities CommissionsIPD Investment Property DatabankIRB Internal-ratings-basedIT Italy IT- ES Italy and SpainJP JapanLCR Liquidity coverage ratioLGD Loss- given- defaultLHS Left- hand sideLIBOR London Inter- Bank Offered RateLIDC Low- income developing countriesLLP Loan loss provisionLLR Loan loss provisions to total assetsLMF Lagrange multiplier testLR Leverage ratioLRS Linear Combination of Ratios of SpacingsLTD Loan- to- deposit ratioLTRO Long- Term Refinancing OperationM Effective maturityM2 M2 money supplyMAS Monetary Authority of Singaporemax MaximumMBS Mortgage- backed securitiesMCR Minimum capital requirementMDB Multilateral development bankMDev Mean absolute deviationMES Marginal expected shortfallmin MinimumMoU Memorandum of UnderstandingMPS Macroprudential policy and surveillanceMSM Minimum solvency margin requirementMtM Mark- to- marketNA North AmericaNAIC National Association of Insurance CommissionersNBB National Bank of BelgiumNBG National Bank of GreeceNI Net incomeNIE Net interest expenseNII Net interest income


Abbreviations xiii

NL NetherlandsNPL Non- performing loanNPR Net profit to total assets ratioNPV Net present valueNSFR Net stable funding ratioNTNI Nontraditional noninsuranceOCC Office of the Comptroller of the CurrencyOCI Other comprehensive incomeOECD Organisation for Economic Co- operation and DevelopmentOeNB Österreichische NationalbankOFR Office of Financial ResearchOIS Overnight indexed swapOLS Ordinary least squaresORSA Own Risk and Solvency AssessmentOSFI Office of the Superintendent of Financial InstitutionsOTC Over- the- counterOW Oliver WymanPB Price- to- bookP&C Property and casualtyPCAR Prudential Capital Assessment ReviewPCR Prescribed capital requirementPD Probability- of- defaultPGI Principal Global IndicatorsPIT Point- in- timeP&L Profit and lossPLAR Prudential Liquidity Assessment ReviewPML Probable maximum lossPRA Prudential Regulation AuthorityPSDS Public Sector Debt StatisticsPSE Public sector entityPT Portugalptb Price to tangible book equity ratioQIS5 Fifth Quantitative Impact StudyR DGI RecommendationRAC Risk- adjusted capitalRAMSI Risk Assessment Model of Systemic InstitutionsRB Retail bankRBC Risk- based capitalRBI Raiffeisen Bank InternationalRCAP Regulatory Consistency Assessment ProgramRDL Royal Decree LawRE Real estateReg RegulatoryRHS Right- hand sideRMBS Residential mortgage- backed securityROC Return on capital


Abbreviationsxiv

RoE Return on equityRPPI Residential Property Price IndicesRTF Research Task ForceRW Risk weightRWA Risk- weighted asset S- 25 Systemic- 25 jurisdictions S- 29 Systemic- 29 jurisdictionsS&P Standard and Poor’sSAD System assets in distressSCAP Supervisory Capital Assessment ProgramSCR Solvency capital requirementSDDS Special Data Dissemination StandardSE SwedenSEB Skandinaviska Enskilda Banken ABSES Systemic expected shortfallSG Slow growthSMEs Small and medium- sized enterprisesSMR Solvency margin ratioSNA System of National AccountsSOE State- owned enterpriseSPV Special purpose vehicleSSM Single Supervisory MechanismSST Swiss Solvency TestST Stress testStA Standardized approachSTeM Stress Test MatrixT1 Tier 1TA Total assetsTARP Trouble Asset Relief ProgramTCE Tangible common equitytce Tangible common equity to total assetsTN Technical NoteTR TurkeyTSR Triennial Surveillance ReviewTTC Through- the- cycleTU Top- downUB Universal bankUCITS Undertakings for Collective Investments in Transferable SecuritiesUK United KingdomUS United StatesVaR Value- at- RiskVIX Chicago Board Options Exchange Volatility IndexWEO World Economic Outlook y- o- y Year- on- yearZCE Zero coupon equivalent



Contributing Authors

IMF Staff (past and present)*

Evrim Bese Goksu, Economist, Statistics Department ([email protected]). Timo Broszeit, Monetary and Capital Markets Department. Currently Independent Expert, Insurance Regulation and Stress

Testing ([email protected]). Elie Canetti, Western Hemisphere Department. Currently Practitioner- in- Residence, Economics and Finance Department, School

of Advanced International Studies, Johns Hopkins University, Washington, DC ([email protected]).Eugenio Cerutti, Deputy Division Chief, Asia and Pacific Department ([email protected]). Jorge A. Chan- Lau, Senior Economist, Strategy, Policy and Review Department ([email protected]). Dimitri G. Demekas, Monetary and Capital Markets Department. Currently Visiting Senior Fellow, Institute of Global Affairs,

London School of Economics and Political Science; and Special Adviser, Bank of England, London, United Kingdom (www. demekas.com).

Dale F. Gray, Monetary and Capital Markets Department. Currently Independent Expert, Financial Stability ([email protected]).

Daniel C. Hardy, Monetary and Capital Markets Department. Currently Academic Visitor at St. Antony’s College, University of Oxford ([email protected]).

Robert Heath, Statistics Department. Currently Fellow, Office for National Statistics, Newport, United Kingdom ([email protected]).

Heiko Hesse, Senior Economist, Strategy, Policy and Review Department ([email protected]).Andreas A. Jobst, Senior Economist, European Department ([email protected]). Kyunghun Kim, Monetary and Capital Markets Department. Currently Assistant Professor, Hongik University, Seoul, Korea

([email protected]). Tidiane Kinda, Senior Economist, Asia and Pacific Department ([email protected]). Emanuel Kopp, Senior Economist, Strategy, Policy and Review Department ([email protected]). Vanessa Le Leslé, Strategy, Policy and Review Department. Currently Independent Consultant ([email protected]). Elena Loukoianova, Deputy Division Chief, Asia Pacific Department ([email protected]). Srobona Mitra, Senior Economist, European Department ([email protected]). Sònia Muñoz, Division Chief, Western Hemisphere Department ([email protected]). Li Lian Ong, Monetary and Capital Markets Department. Currently Group Head, Financial Surveillance, ASEAN+3 Macroeco-

nomic Research Office, Singapore ([email protected]). Hiroko Oura, Deputy Division Chief, Monetary and Capital Markets Department ([email protected]). Ceyla Pazarbasioglu, Director, Strategy, Policy, and Review Department ([email protected]).Ferhan Salman, Senior Economist, Middle East and Central Asia Department ([email protected]). Christine Sampic, Monetary and Capital Markets Department. Currently Deputy Director General, Banknote Manufacturing,

Banque de France, Paris, France ([email protected]). Christian Schmieder, Monetary and Capital Markets Department. Currently Head of Administration, Monetary and Economic

Department, Bank for International Settlements, Basel, Switzerland ([email protected]).Liliana Schumacher, Senior Economist, Monetary and Capital Markets Department ([email protected]). Nobuyasu Sugimoto, Senior Financial Sector Expert, Monetary and Capital Markets Department ([email protected]). Laura Valderrama, Senior Economist, European Department ([email protected]). Francis Vitek, Senior Economist, Monetary and Capital Markets Department ([email protected]). Rodolfo Wehrhahn, Monetary and Capital Markets Department. Currently International Consultant, Insurance and Pensions

([email protected]).

*Chapters were written while all authors were staff at the IMF.

http://www.demekas.com

http://www.demekas.com

Contributing Authorsxvi

External Coauthors

Stefan W. Schmitz, Head of Macroprudential Supervision, Financial Stability and Macroprudential Supervision Division, Oesterreichische Nationalbank ([email protected]).

Michael Sigmund, Senior Analyst, Financial Stability and Macroprudential Supervision Division, Oesterreichische National-bank ([email protected]).

Nassim Nicholas Taleb, Universa Investments, Scientific Adviser ([email protected]).

Disclaimer

The views expressed in this volume are those of the authors and do not necessarily represent those of their respective institutions.


CHAPTER 1

Stress Testing at the International Monetary Fund: Principles, Concepts, and Frameworks

ANDREAS A. JOBST • LI LIAN ONG

illustrate the anatomy of IMF stress testing (Figure 1.1) and discuss several areas where the IMF staff has made progress, notably: (1) the design of and model selection for stress tests; (2) key gaps in the coverage of risks and information, which were highlighted during the GFC; (3) practical approaches to implementing important aspects of stress testing; and (4) widening the scope of financial activities beyond banks, namely, nonbank financial institutions and financial market infrastructures (FMIs).

That said, this book does not claim to be comprehensive, and much remains to be done on the stress testing front as an active area of work for the IMF. It does not capture the work- in- progress owing to the timing of its production, such as the IMF staff’s work on ongoing financial stability assess-ments and related implications of the current COVID-19 pandemic as well as the 2020 Review of the FSAP.1 The em-pirical analysis underlying these chapters has not been sub-stantially updated since the issuance of the respective working papers. However, references have been updated throughout to reflect important regulatory and research de-velopments. In all, the chapters in this volume provide help-ful insights into the IMF staff’s evolving thinking on the principles, concepts, and frameworks of stress testing over the past eight years.

2. ANATOMY OF STRESS TESTINGEssential Building Blocks

“Best practice” principles provide a baseline against which individual stress testing exercises may be assessed. In Chapter 2, Schumacher and Oura (with others) discuss the

1 More recent studies, such as those by Krznar and Matheson (2017) and Bouveret (2018), have not been included.

1. MOTIVATIONStress testing at the IMF has evolved into an integral aspect of financial sector surveillance over the past two decades. It is a key component of the Financial Sector Assessment Pro-gram (FSAP) and was used during the global financial crisis (GFC) to support estimations of banking system recapital-ization needs of countries with the IMF’s crisis programs. Stress testing has also become an important forward- looking riskmanagement tool for financial supervisors and macro-prudential authorities to identify vulnerabilities of individ-ual banks and financial systems to the impact of adverse changes to the operational and market environment.

The intensified interest in stress testing has underscored the need for developing and communicating a coherent and consistent approach for such exercises as part of the IMF’s financial surveillance. The first book, A Guide to IMF Stress Testing: Methods and Models (which was published in 2014), focuses predominantly on the stress testing models devel-oped by the IMF staff for financial stability analysis and cri-sis management. It categorizes relevant analytical tools and methods into three distinct but complementary approaches (that is, accounting- based, market- price- based, and macro- financial). It also showcases continuing efforts by the IMF staff to apply more encompassing risk measures and assess-ment methods and summarizes their advantages and disad-vantages in recommending their appropriate application.

The development of comprehensive stress testing policies has facilitated better harmonization and comparability of approaches, methods, and models. This anthology comple-ments the first book by compiling selected papers written by several IMF staff members (and their external coauthors) on principles, concepts, and frameworks of stress testing. They


2Stress Testing at the International M

onetary Fund: Principles, Concepts, and Framew

orks

Source: Authors.1Financial auxiliaries are financial corporations that are principally engaged in activities associated with transactions in financial assets and liabilities or with providing the regulatory context for these transactions, which do not involve the auxiliary taking ownership of the financial assets and liabilities being transacted according to the taxonomy of the Global Monitoring Report on Non-bank Financial Intermediation (FSB 2019).2Other financial intermediaries include captive financial institutions and money lenders, investment funds, trust companies, central counterparties, broker-dealers, and securitization vehicles.

Figure 1.1. Anatomy of Stress Testing: Principles, Concepts, and Frameworks

Bank Stress Testing

Macroscenarios

Nonbank Stress Testing

IMF STRESS TESTING

Solvency: Frameworkand Application (Ch. 15)

Interaction of Solvencyand Funding Costs (Ch. 8)

Financial Information(Consolidated/Solo)

Impact on Liquidity

First Book: Methods &Models

Impact on Solvency

First Book: Methods &Models

Market and FundingScenarios

Macroshocks and Feedback Effectsto Solvency and Liquidity (Ch. 12)

Information Gaps (Ch. 5)

Ring-Fencing, RegulatoryDifferences (Ch. 6)

Cross-Border Credit andFunding Shocks (Ch. 13)

Crisis (vs. Peacetime)Solvency Stress Testing (Ch. 14)

Liquidity: Frameworkand Application (Ch. 16)

Principles and Practices (Ch. 2)

Financial Information(Consolidated/Solo)

Scenario Simulation(Ch. 4)

Other

Risk-Weighted Assets(Ch. 10)

Pension Funds(First Book: Methods & Models)

Financial Auxiliaries1

Public Financial Institutions

Other FinancialIntermediaries2

Credit Rules of Thumb (Ch. 7) Sovereign Haircuts (Ch. 9)

Satellite Models(First Book: Methods & Models)

Tail Risk Heuristic (Ch. 11)

Selection/Design of Models and Other Elements (Ch. 3)

Insurance(Ch. 17)

Islamic Banks

Nonfinancial Firms Stress Testing

Other Concepts/Tools


Andreas A. Jobst and Li Lian Ong 3

tically or abroad, with the banking sector exposed to them either directly, via determinants of profitability and capi-talization, or indirectly through macro- financial linkages to, between, or among these determinants. However, the consistent calibration of shocks that affect the interrela-tionships across various macro- financial variables tends to be challenging.

The difficulty in designing scenarios increases if the shocks are assumed to reverberate across several countries, affecting financial institutions that have significant cross- border activities. Vitek addresses this issue in Chapter 4, us-ing a structural macroeconomic model of the world economy, which features a range of nominal and real rigidities, exten-sive macro- financial linkages, and diverse spillover transmis-sion channels. More recently, the growth- at- risk framework was used to inform stress testing scenario design based on how financial conditions, sectoral imbalances, and external factors influence the tail risks to future GDP growth (IMF 2017f; Prasad and others 2019). This concept was applied in the specification of both the baseline and adverse scenarios over a three- year risk horizon in the context of the Peru FSAP (IMF 2018c).

Quality information is essential for useful and credible stress tests. In Chapter 5, Heath and Bese Goksu, in a 2016 paper, observe that the availability of “right data” might have made it possible to better detect risk buildups prior to the GFC (even though the lack of data was not the main cause of the crisis). Information gaps could create additional un-certainty about the performance and soundness of financial institutions during periods of systemic stress. While pruden-tial data are generally available to national supervisors, they remain confidential and often cannot be shared with super-visors in host countries, much less disclosed to the general public. This may create challenges for country authorities who may want to assure investors and depositors of the health of financial institutions during times of stress when trust in the financial system is critical. There are several key examples of how information gaps have undermined market confidence during the GFC:

• The variation and limited transparency in the calcu-lation of risk- weighted assets across banks and juris-dictions created uncertainty about banks’ capital adequacy. This issue is covered by Le Leslé and Jobst in Chapter 10.

• The crisis stress tests conducted by the European au-thorities did not disclose important information on banks’ sovereign risk exposures, which was widely regarded as a major vulnerability. This issue is dis-cussed by Ong and Pazarbasioglu in Chapter 14.

Data limitations may also require reliance on rules of thumb for stress tests. In Chapter 7, Hardy and Schmieder show that rules of thumb may have to be imported in the design of stress tests for a country in situations such as when national authorities or bank management are unable to estimate behavioral relationships robustly based on available data.

practical guidelines derived from the IMF staff’s stress test-ing experience up to 2012 and a survey of stress testing practices among selected national central banks and super-visory authorities conducted in 2011. The authors argue that overarching principles can effectively guide many of the choices in designing and implementing system- wide stress tests. These areas include (1) the coverage of institu-tions, risks, and scenarios; (2) the specification of a suitable quantitative framework to link various shock scenarios to solvency and liquidity measures; (3) a strategy for commu-nicating the results; and (4) follow- up measures, if war-ranted.2 These considerations complement the updated supervisory principles published by the Basel Committee on the use, implementation, and oversight of stress testing frameworks regarding their objectives, governance, poli-cies, processes, methodology, resources, and documenta-tion (BCBS 2018).

The implementation of stress testing principles and con-cepts for macroprudential surveillance and financial stability analysis is closely intertwined with the development of sys-temic risk models. In Chapter 3, Demekas discusses the re-quired features of tools that can effectively assess system- wide risks; notably, they would have to (1) incorporate general equilibrium dimensions, and (2) focus on the resilience of the system as a whole. However, while stress testers have made significant progress in the first area (some of which are covered in this book), they have achieved much less in ad-dressing the second.

The effective implementation of stress tests requires frameworks that adequately cover both solvency and liquid-ity risks. Within the banking sector, the focus had tradition-ally been on solvency risk, but liquidity risk came to the fore at the onset of the GFC. In Chapters 15 and 16, Jobst, Ong, and Schmieder illustrate the application of stress testing principles and concepts by reviewing key elements of the IMF staff’s solvency and liquidity stress tests and cataloging their actual implementation in FSAPs for jurisdictions with systemically important financial sectors between 2010 and 2016. While a similar solvency stress testing framework may be used for crisis stress tests, Ong and Pazarbasioglu under-score in Chapter 14 that the design of some of those ele-ments is invariably different to ensure the credibility of the exercise at a most critical time.

Another important aspect of stress testing is the simula-tion of shock scenarios. Banking sector solvency stress tests typically apply adverse— that is, severe but plausible— macro- financial scenarios. These scenarios describe the evolution of various macro- financial variables over a spe-cific horizon, following shocks that move them away from their respective baselines. The shocks may originate domes-

2 The design of a stress test and the communication of its results should be fully aligned with the policy objectives and the applicable restructur-ing and resolution regime, which Baudino and others (2018) confirm in their review of the three main building blocks (governance, implemen-tation, and outcomes) of system- wide bank stress tests in the euro area, Japan, Switzerland, and the United States.


Stress Testing at the International Monetary Fund: Principles, Concepts, and Frameworks4

failure to incorporate the solvency- liquidity nexus in stress tests could lead to significant underestimation of shocks on bank capitalization. They also find evidence of nonlinearity between solvency and funding costs.

The practical application of the solvency- liquidity linkage in stress tests remains at an early stage. A limited set of exist-ing stress testing models contain liquidity and solvency in-teractions or network modules, in considering contagion and systemic risk from a cross- functional perspective (Barn-hill and Schumacher 2011; Schmieder and others 2012; Babihuga and Spaltro 2014; Jobst 2014; Krznar and Mathe-son 2017; Cont, Kotlicki, and Valderrama 2019). While these models have remained necessarily simple, given vari-ous empirical and data constraints, they could be extended to accommodate prudential data, which would help refine scenario specifications such as the duration of shocks and the adjustment dynamics as the financial system stabilizes over time ( Grillet- Aubert 2018). They could also be inte-grated with relevant analytical tools, such as general equilib-rium models, agent- based models, networks, and behavioral analysis.

Feedback loops with the real economy

The two- way interaction between the real economy and fi-nancial activities, and related feedback effects generated by financial institutions’ reaction function to stress, requires a dynamic specification of transmission channels. It also ne-cessitates the consistent and comparable design of macro- financial scenarios. Such specifications could be enriched with insights into the adjustment process of economic agents to price and output shocks from full (or partial) equilibrium macroeconomic models. Work in this area remains at an early stage, with the IMF staff developing models that can credibly capture important feedback effects. In Chapter 12, Hesse, Ferhan, and Schmieder simulate the potential impact of spillovers from a crisis on banks’ liquidity and capital po-sitions, and then examine their impact on the real economy. They find that spillovers have a highly nonlinear impact on both aspects (liquidity and solvency) of bank soundness.

Spillover effects from interconnectedness

The interlinkages within financial systems (such as inter-bank markets) or interactions between financial and nonfi-nancial entities within and across national boundaries (through common exposures, such as the property market) can result in financial contagion and spillovers. Prior to the GFC, credit risk shocks in FSAP stress tests focused largely on the capital impact of banks’ local exposures to firms and households without considering cross- border exposures (through branches and subsidiaries). Since then, spillover ef-fects have been introduced through network analysis, for example, in the FSAPs for Australia, France, Japan, Spain, and the United States (IMF 2012a, 2012b, 2013a, 2017e, 2019); the simulation of ring- fencing, for example, the Spain FSAP (IMF 2012b); and shocks to business activities in

Addressing Gaps in Risk Coverage and Assessment

FSAP stress tests attempt to cover the relevant sources of risk affecting capital and liquidity conditions in the financial sys-tem in adverse macro- financial scenarios. The outcomes of stress tests are driven by the initial identification of these risks— in detecting, monitoring, and mitigating their buildup based on known vulnerabilities from common ex-posures, risk concentrations, and interdependencies within the financial system. The IMF staff has made significant ef-forts to close important gaps in risk coverage that were high-lighted by the GFC to ensure that FSAP stress tests are fit for this purpose and encompass the following four essential do-mains, namely: (1) the dynamic approach to modeling insti-tutional behavior under stress, (2) the interaction between solvency and liquidity risks, (3) feedback loops with the real economy, and (4) spillover effects from interconnectedness.

Dynamic approach to modeling institutional behavior under stress

A more dynamic approach considers changes in institutional behavior that can impact both capital and liquidity condi-tions under adverse macro- financial scenarios. In stress situ-ations, banks may ring- fence liquidity, adjust their balance sheets, and/or restrict profit distribution, or be required to do so. These measures could include (1) limiting flows of funds within a banking group, (2) increasing liquidity through as-set sales and/or slowing credit growth, (3) raising capital, and/or (4) reducing dividends. Hence, stress tests should adopt a more dynamic approach that considers changes in regulatory requirements or institutional behavior that could affect banks’ financial statements. In Chapter 6, Cerutti and Schmieder show that the use of both consolidated and un-consolidated balance sheet data is necessary to estimate the potential impact of ring- fencing on international banking groups. This dynamic approach was included as part of the bank stress tests in the Spain FSAP, which took place during the European sovereign debt crisis (IMF 2012b).

Interaction between solvency and liquidity risks

Liquidity and solvency risks faced by individual institutions are increasingly intertwined during times of stress. They tend to be influenced by system- wide liquidity conditions associated with the interconnectedness and network effects within the financial system. Stress tests that do not account for the interaction between solvency and liquidity shocks substantially underestimate the risk exposure of individual banks and banking sectors (Puhr and Schmitz 2014; BCBS 2015). Empirical evidence suggests that the interaction be-tween solvency and funding costs (1) is indeed statistically significant and (2) might be economically relevant, espe-cially during periods of stress. In Chapter 8, Schmitz, Sig-mund, and Valderrama show that the interactions between liquidity and solvency risks are material and argue that



Wider Coverage of Financial Activities

The increasing importance of nonbank financial services for macroprudential surveillance has gradually widened the pe-rimeter of stress testing beyond the banking sector. Im-proved data availability, enhanced statutory reporting, and supervisory coordination facilitate the integration of a wider range of nonbank financial institutions and markets into stress tests. However, limited data access and various other data- related factors (such as standards and classifications) continue to constrain the development of comprehensive stress testing models and frameworks for non bank financial intermediaries and auxiliaries.

Building a complete map of funding interconnections be-tween money and derivatives markets and various financial entities remains challenging. While a macroprudential per-spective to stress testing and regulations could have helped prevent the GFC, Aikman and others (2019) argue that it would still have been difficult to understand the fragility of funding flows across the system and their knock- on effects on nonbank financial entities without covering the entire fi-nancial system to reveal the full extent of existing vulnera-bilities. Clearly, there are still significant gaps in the coverage of all relevant market participants and the interlinkages be-tween solvency and liquidity conditions across banks and important FMI elements.

At the IMF, stress testing of the insurance sector has be-come a more regular exercise during FSAPs since the GFC. Indeed, the IMF staff has developed a common approach to stress testing insurance companies. In Chapter 17, Jobst, Sugimoto, and Broszeit review the state of system- wide sol-vency stress tests for insurers, comparing national practices and drawing on experience from FSAPs to derive principles and concepts, and distill practical guidelines for a more comparable and consistent implementation of system- wide insurance stress tests.

The IMF staff has also run stress tests on other nonbank financial institutions. They include pension funds, invest-ment funds, and critical FMI elements, including central counterparties (CCPs) and central securities depositories. The technical guidelines for stress testing defined benefit pension plans are covered in the first book. Technical work on liquidity stress testing for investment funds and its poten-tial integration with bank stress tests was undertaken in the context of the FSAPs for Ireland, Luxembourg, Sweden, and the United States (Bouveret 2017; IMF 2015b, 2016e, 2017c, 2017d). Detailed risk assessments of critical FMI elements have been undertaken in several FSAPs, focusing on the largest central counterparties and central securities deposito-ries in Europe and the United States.3 The financial stability analysis of CCPs varies across countries and is largely driven

3 For instance, Euroclear (Belgium), Eurex (Germany), CC&G (Italy), Clearstream (Luxembourg), Euro CCP (Netherlands), Nasdaq (Swe-den), LCH (United Kingdom), and CME (United States) (IMF 2013a, 2013b, 2013d, 2015a, 2016a, 2016d, 2016f, 2017a, 2017b, 2017d, 2018a).

other countries, for example, the United Kingdom FSAP (IMF 2016c). In Chapter 13, Kim and Mitra explore the sig-nificance of spillover effects from cross- border banking link-ages. They use the network model of Espinosa- Vega and Solé (2011) to estimate the impact of credit risk and funding shocks on bank capital through direct and indirect trans-mission channels. They also model the relationship between spillover effects from cross- border credit and funding shocks and GDP growth rate surprises and find that funding vul-nerabilities have implications mostly for the latter.

Other facets of solvency risk

Solvency stress tests remain largely focused on credit and mar-ket risks. The most common among the latter are interest rates, foreign exchange rates, and credit spreads, as well as equity, real estate, and commodity prices. However, the GFC high-lighted additional facets of solvency risk, which have since been included in FSAP stress tests, notably: (1) the nonlinear impact of solvency on funding costs, which is demonstrated by Schmitz, Sigmund, and Valderrama in Chapter 8; (2) the po-tential valuation losses of sovereign exposures and other low- default assets, which have influenced stress tests via the analysis of the bank- sovereign nexus (IMF 2018b, 2018d) and are cov-ered by Jobst and Oura in Chapter 9; and (3) the realization of contingent liabilities (such as guarantees, commitments, and derivatives), which have been mostly addressed via the impact of net cash outflows on the capital assessment of banks under stress in the United Kingdom FSAP (IMF 2016b).

Practical Approaches in the Implementation of Important Conceptual Aspects of Stress Testing

Stress testing is arguably more art than science. While the stress test methods and models covered in the first book are technical in nature, the principles, concepts, and frameworks described in this book often require some expert judgment and experience, especially if policy considerations or empiri-cal constraints require practical and pragmatic solutions at different stages of a stress testing exercise. Heuristics can facilitate the calibration of shocks and the modeling of typi-cal behavioral relationships, complemented by detailed anal-yses of banks’ financial statements and circumstances. In Chapter 7, Hardy and Schmieder identify several helpful rules of thumb for stress testing bank solvency, with a focus on the elasticity of credit losses, preimpairment income, and credit growth during times of stress.

Stress test results could also be influenced by model er-rors and parameter uncertainty. In Chapter 11, Taleb and others introduce a simple approach that helps evaluate how well tail risks are captured in stress tests. The authors argue that stress tests capture only first- order effects of negative impacts associated with tail shocks; they propose a heuristic as a second- order (robustness) test to detect nonlinearities in the tail behavior of risks. More specifically, their method identifies potential convexities that could under- or overstate the impact of tail events.



Liquidity and Solvency Interactions and Systemic Risk.” BIS Working Paper No. 29, Bank for International Settlements, Basel. https://www.bis.org/bcbs/publ/wp29.htm.

———. 2018. “Stress Testing Principles.” Guidelines, BIS Paper No. 450, Bank for International Settlements. https://www.bis .org/bcbs/publ/d450.htm.

Baudino, Patrizia, Roland Goetschmann, Jérôme Henry, Ken Taniguchi, and Weisha Zhu. 2018. “Stress- Testing Banks— A Comparative Analysis.” FSI Papers No. 12, Bank for Interna-tional Settlements, Basel. https://www.bis.org/fsi/publ/insights12 .htm.

Bouveret, Antoine. 2017. “Liquidity Stress Tests for Investment Funds: A Practical Guide.” IMF Working Paper 17/226, Inter-national Monetary Fund, Washington, DC. https://www.imf .org/en/Publications/WP/Issues/2017/10/31/ Liquidity -Stress-Tests- for- Investment-Funds-A-Practical-Guide-45332.

———. 2018. “Cyber Risk for the Financial Sector: A Framework for Quantitative Assessment.” IMF Working Paper 18/143, In-ternational Monetary Fund, Washington, DC. https://www.imf .org/en/Publications/WP/Issues/2018/06/22/Cyber-Risk -for-the-Financial-Sector-A-Framework-for-Quantitative -Assessment-45924.

Cont, Rama, Arur Kotlicki, and Laura Valderrama. 2019. “Liquid-ity at Risk: Joint Stress Testing of Solvency and Liquidity.” SSRN Working Paper, June. https://ssrn.com/abstract=3397389.

Espinosa- Vega, Marco, and Juan Solé. 2011. “ Cross- Border Finan-cial Surveillance: A Network Perspective.” Journal of Financial Economic Policy 3 (3): 82−205.

European Securities and Markets Authority (ESMA). 2018. “Report: EU- wide CCP Stress Test 2017.” ESMA, Paris. https://www.esma.europa.eu/press-news/esma-news/esma-publishes -results-second-eu-wide-ccp-stress-test.

Financial Stability Board (FSB). 2019. “Global Monitoring Report on Non- Bank Financial Intermediation 2018.” Bank for Inter-national Settlements, Basel, February. http://www.fsb.org/2019 /02/ g loba l- monitoring- report- on-non-bank-f inancia l -intermediation-2018/.

Grillet- Aubert, Laurent. 2018. “Macro Stress Tests: What Do They Mean for Market and for the Asset Management Industry?” Working Paper, Research, Strategy and Risks Directorate, Autorité des Marchés Financiers (AMF), Paris, June. https:// www.amf- france.org/en_US/Publications/ Lettres- et- cahiers /Risques-et-tendances/Archives?docId=workspace%3A%2F%2FSpacesStore%2F28bd5080-6c2d-4154-a015-1b7678210b64.

He, Dong, Ross B. Leckow, Vikram Haksar, Tommaso Mancini- Griffoli, Nigel Jenkinson, Mikari Kashima, Tanai Khiaonar-ong, Celine Rochon, and Hervé Tourpe. 2017. “Fintech and Financial Services: Initial Considerations.” IMF Staff Discus-sion Notes 17/05, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/Staff-Discussion -Notes/Issues/2017/06/16/Fintech-and-Financial-Services -Initial-Considerations-44985.

International Monetary Fund (IMF). 2012a. “Japan: Financial Sector Assessment Program: Technical Note on Financial Sys-tem Spillovers–An Analysis of Potential Channels.” IMF Country Report 12/263, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/CR/Issues /2016/12/31/ Japan- Financial- Sector- Assessment- Program - Technical-Note-on-Financial-System-Spillovers-An-26247.

———. 2012b. “Spain: Financial System Stability Assessment.” IMF Country Report 12/137, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR

by existing stress testing frameworks developed by local reg-ulators (Anderson, Cerezetti, and Manning 2018; ESMA 2018). The FSAPs encouraged national authorities to de-velop standardized stress testing to support their assessment of CCPs’ loss- absorbing capacity.

The nature of stress testing continues to expand. Its scope is becoming more diverse to account for different characteris-tics of financial services across countries. For instance, the IMF staff has been developing conceptual guidelines for the coherent implementation of solvency stress testing of Islamic banks. The formulation of stress tests also considers the evolving nature of risks. For example, new types of risks are emerging, such as climate change and technological disruptions. The IMF staff has provided initial considerations on these risks (He and others 2017; Jobst and Pazarbasioglu 2019), but a fuller discussion on their incorporation into stress tests is outside the coverage of this book.

Stress testing exercises comprise many “moving parts” of continuously evolving risks and the methodologies required to adequately capture them. Managing these dynamics is even more challenging in the context of FSAPs, which need to be sufficiently comprehensive and comparable for a wide range of countries with varying characteristics of financial systems. FSAP stress tests must also remain sufficiently flexible to accommodate the specific nature of different local regulatory requirements (including on data) and political sensitivities. Hence, the development of common principles and concepts for well- designed stress tests, all within coher-ent frameworks, ensures that the financial stability analysis in FSAPs remains relevant and appropriate while comple-menting the efforts of national authorities in developing their own stress tests.

REFERENCESAikman, David, Jonathan Bridges, Anil Kashyap, and Caspar

Siegert. 2019. “Would Macroprudential Regulation Have Prevented the Last Crisis?” Journal of Economic Perspectives 33 (1): 107–30.

Anderson, Edward, Fernando Cerezetti, and Mark Manning. 2018. “Supervisory Stress Testing for CCPs: A Macro- Prudential, Two- Tier Approach.” Finance and Economics Dis-cussion Series 2018-082, Board of Governors of the Federal Reserve System, Washington, DC.

Barnhill, Theodore, and Liliana Schumacher. 2011. “Modeling Correlated Systemic Liquidity and Solvency Risks in a Finan-cial Environment with Incomplete Information.” IMF Work-ing Paper 11/263, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/ 12/31/Model ing-Corre lated-Systemic-L iqu id it y-and -Solvency- Risks- in-a-Financial-Environment-with-25356.

Babihuga, Rita, and Marco Spaltro. 2014. “Bank Funding Costs for International Banks.” IMF Working Paper 14/71, International Monetary Fund, Washington, DC. https://www.imf.org/en /Publications/WP/Issues/2016/12/31/ Bank-Funding-Costs-for -International-Banks-41514.

Basel Committee on Banking Supervision (BCBS). 2015. “Making Supervisory Stress Tests More Macroprudential: Considering


https://www.bis.org/bcbs/publ/wp29.htm.

https://www.bis.org/bcbs/publ/d450.htm


https://www.bis.org/fsi/publ/insights12.htm

https://www.bis.org/fsi/publ/insights12.htm

https://www.imf.org/en/Publications/WP/Issues/2017/10/31/


https://www.imf.org/en/Publications/WP/Issues/2018/06/22/Cyber-Risk-for-the-Financial-Sector-A-Framework-for-Quantitative-Assessment-45924




https://ssrn.com/abstract=3397389

https://www.esma.europa.eu/press-news/esma-news/esma-publishes-results-second-eu-wide-ccp-stress-test



https://www.imf.org/en/Publications/Staff-Discussion-Notes/Issues/2017/06/16/Fintech-and-Financial-Services-Initial-Considerations-44985



https://www.imf.org/en/Publications/CR

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Modeling-Correlated-Systemic-Liquidity-and-Solvency-





www.amf- france.org/en_US/Publications/ Lettres- et- cahiers/Risques-et-tendances/Archives?docId=workspace%3A%2F%2FSpacesStore%2F28bd5080-6c2d-4154-a015-1b7678210b64.

www.amf- france.org/en_US/Publications/ Lettres- et- cahiers




http://www.fsb.org/2019/02/global-monitoring-report-on-non-bank-financial-intermediation-2018



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Japan-Financial-Sector-Assessment-Program-Technical-Note-on-Financial-System-Spillovers-An-26247




-Financial-Sector-Assessment-Program-Systemic-Risk-and -Interconnectedness-43975.

———. 2016d. “Germany: Financial Sector Assessment Program —Detailed Assessment of Observance on the Eurex Clearing AG Observance of the CPSS- IOSCO Principles for Financial Market Infrastructures.” IMF Country Report 16/197, Interna-tional Monetary Fund, Washington, DC. https://www.imf .org/en/Publicat ions/CR /Issues/2016/12/31/Germany - F i n a nc i a l - S e c to r- A s s e s sment- P rog r a m- D e t a i l e d - Assessment- of-Observance-on-the-Eurex-44021.

———. 2016e. “Ireland: Financial Sector Assessment Program— Technical Note: Asset Management and Financial Stability.” IMF Country Report 16/312, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/Ireland-Financial-Sector-Assessment-Program -Technical- Note-Asset-Management-and-Financial-44305.

———. 2016f. “Sweden: Financial System Stability Assessment.” IMF Country Report 16/355, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/ Sweden-Financia l-System-Stability -Assessment-44404.

———. 2017a. “Kingdom of the Netherlands— Netherlands: Financial Sector Assessment Program: Technical Note— Regulation, Supervision, and Oversight of Financial Market Infrastructures— Responsibilities and EuroCCP Financial and Operational Risk Management.” IMF Country Report 17/92, International Monetary Fund, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2017/04/13/Kingdom -of-the-Netherlands-Netherlands-Financial-Sector-Assessment -Program-Technical-Note-44817.

———. 2017b. “Luxembourg: Technical Note— Detailed Assess-ment of Observance— Assessment of Observance of the CPSS- IOSCO Principles for Financial Market Infrastructures: Clearstream Banking.” IMF Country Report 17/260, Interna-tional Monetary Fund, Washington, DC. https://www.imf.org /en/Publications/CR/Issues/2017/08/28/Luxembourg -Financial-Sector-Assessment-Program-Detailed-Assessment -of-Observance-Assessment-45209.

———. 2017c. “Luxembourg Financial Sector Assessment Pro-gram: Technical Note— Risk Analysis.” IMF Country Report 17/261, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2017/08/28 /Lu xembourg-Fina nc ia l -Sec tor-A s se s smentProg ra m -Technical-Note-Risk-Analysis-45210.

———. 2017d. “Sweden: Financial Sector Assessment Program Technical Note— Supervision and Oversight of Financial Mar-ket Infrastructures.” IMF Country Report 17/310, Interna-tional Monetary Fund, Washington, DC. https://www.imf.org /en/Publications/CR/Issues/2017/10/05/Sweden-Financial -Sector-Assessment-Program-Technical-Note-Supervision -and-Oversight-of-45304.

———. 2017e. “Spain: Financial Sector Assessment Program— Technical Note— Interconnectedness and Spillover Analysis in Spain’s Financial System.” IMF Country Report 17/344, Inter-national Monetary Fund, Washington, DC. https://www.imf.org /en/Publications/CR/Issues/2017/11/13/Spain-Financial-Sector -Assessment-Program-Technical-Note-Interconnectedness -and-Spillover-45395.

———. 2017f. Global Financial Stability Report: Is Growth at Risk? Chapter 1. Washington, DC, October. https://www.imf.org /en/Publications/GFSR/Issues/2017/09/27/global-financial -stability-report-october-2017.

/Issues/2016/12/31/ Spa in-Financia l-System-Stabi l it y -Assessment-25977.

——— 2013a. “France: Financial Sector Assessment Program— Technical Note on Stress Testing the Banking Sector.” IMF Country Report 13/185, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/CR/Issues /2 016 /12 /31/ Fr a nc e -F i n a nc i a l - S e c to r-A s s e s s me nt -Program-Technical-Note- on-Stress-Testing-the-Banking -40722.

———. 2013b. “Euro Area Policies: Technical Note— Supervision and Oversight of Central Counterparties and Central Securi-ties Depositories.” IMF Country Report 18/227, International Monetary Fund, Washington, DC. https://www.imf.org/en /Publications/CR/Issues/2018/07/19/Euro-Area-Policies -Financia l-Sector- Assessment- Program-Technica l-Note -Supervision-and-46101.

———. 2013c. “European Union: Detailed Assessment of Obser-vance of the CPSS- IOSCO Principles for Financial Market In-frastructures.” IMF Country Report 13/332, International Monetary Fund, Washington, DC. https://www.imf.org/en /Publicat ions/CR /Issues/2016/12/31/European-Union -Publicat ion-of-Financia l- Sector-Assessment-Program -Documentation-Detailed-41068.

———. 2013d. “Italy: Technical Note— Financial Risk Manage-ment and Supervision of Cassa Di Compensazione e Garanzia S.P.A.” IMF Country Report 13/351, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /CR/Issues/2016/12/31/Italy-Technical-Note-on-Financial -R i s k- M a n a g e m e nt- a nd - Sup e r v i s i on - o f - C a s s a -D i -Compensazione-41092.

———. 2015a. “United States: Financial Sector Assessment Pro-gram Technical Note— Detailed Assessment of Implementa-tion on the IOSCO Objectives and Principles of Securities Regulation.” IMF Country Report 15/91, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /CR/Issues/2016/12/31/United-States-Financia l-Sector -Assessment-Program- Detailed-Assessment-of-Implementation -on-42827.

———. 2015b. “United States: Financial Sector Assessment Pro-gram Technical Note— Stress Testing.” IMF Country Report 15/173, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12 /31/ United- States- Financial- Sector- Assessment- Program -Stress-Testing-Technical-Notes-43058.

——— 2016a. “United Kingdom: Financial Sector Assessment Program— Supervision and Systemic Risk Management of Fi-nancial Market Infrastructures: Technical Note.” IMF Coun-try Report 16/156, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12 /31/United-Kingdom-Financial-Sector-Assessment-Program - Supervision-and-Systemic-Risk-Management-43967.

———. 2016b. “United Kingdom: Financial Sector Assessment Program— Stress Testing the Banking Sector: Technical Note.” IMF Country Report 16/163, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/United-Kingdom-Financial-Sector-Assessment -Program- Stress-Testing-the-Banking-Sector-43974.

———. 2016c. “United Kingdom: Financial Sector Assessment Program— Systemic Risk and Interconnectedness Analysis— Technical Note.” IMF Country Report 16/164, International Monetary Fund, Washington, DC. https://www.imf.org/en /Publications/CR/Issues/2016/12/31/United-Kingdom


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Germany-


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Ireland-Financial-Sector-Assessment-Program-Technical-Note-Asset-Management-and-Financial-44305.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/


https://www.imf.org/en/Publications/CR/Issues/2017/04/13/Kingdom-of-the-Netherlands-Netherlands-Financial-Sector-Assessment-Program-Technical-Note-44817.




https://www.imf.org/en/Publications/CR/Issues/2017/08/28/Luxembourg-Financial-Sector-Assessment-Program-Detailed-Assessment-of-Observance-Assessment-45209.




https://www.imf.org/en/Publications/CR/Issues/2017/08/28/Luxembourg-Financial-Sector-AssessmentProgram-Technical-Note-Risk-Analysis-45210.



https://www.imf.org/en/Publications/CR/Issues/2017/10/05/Sweden-Financial-Sector-Assessment-Program-Technical-Note-Supervision-and-Oversight-of-45304.




https://www.imf.org/en/Publications/CR/Issues/2017/11/13/Spain-Financial-Sector-Assessment-Program-Technical-Note-Interconnectedness-and-Spillover-45395.




https://www.imf.org/en/Publications/GFSR/Issues/2017/09/27/global-financial-stability-report-october-2017./Issues/2016/12/31/




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/France-Financial-Sector-Assessment-Program-Technical-Note-on-Stress-Testing-the-Banking-40722.




https://www.imf.org/en/Publications/CR/Issues/2018/07/19/Euro-Area-Policies-Financial-Sector-Assessment-



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/European-Union-Publication-of-Financial-Sector-Assessment-Program-Documentation-Detailed-41068.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Italy-Technical-Note-on-Financial-Risk-Management-and-Supervision-of-Cassa-Di-Compensazione-41092.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-States-Financial-Sector-Assessment-Program-Detailed-Assessment-of-Implementation-on-42827.





https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-Kingdom-Financial-Sector-Assessment-Program-


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-Kingdom-Financial-Sector-Assessment-Program-Stress-Testing-the-Banking-Sector-43974.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-Kingdom

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-Kingdom










Jobst, Andreas A. 2014. “Measuring Systemic Risk- Adjusted Li-quidity (SRL)—A Model Approach.” Journal of Banking and Finance 45 (C): 270–87.

———, and Ceyla Pazarbasioglu. 2019. “Greater Transparency and Better Policy for Climate Finance.” Financial Stability Review. Banque de France, Paris, June: 85–100.

Krznar, Ivo, and Troy Matheson. 2017. “Towards Macroprudential Stress Testing: Incorporating Macro Feedback Effects.” IMF Working Paper 17/149, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues / 2 017/ 0 6 / 3 0 / To w a r d s - M a c r o p r u d e n t i a l - S t r e s s - Testing-Incorporating-Macro-Feedback-Effects-44955.

Prasad, Ananthakrishnan, Selim Elekdag, Phakawa Jeasakul, Ro-main Lafarguette, Adrian Alter, Alan Xiaochen Feng, and Changchun Wang. 2019. “Growth at Risk: Concept and Ap-plication in IMF Country Surveillance.” IMF Working Paper 19/36, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2019/02/21/ Growth - at- R isk- Concept- and- Appl icat ion-in-IMF-Countr y -Surveillance-46567.

Puhr, Claus, and Stefan W. Schmitz. 2014. “A View from the Top— The Interaction Between Solvency and Liquidity Stress.” Journal of Risk Management in Financial Institutions 7 (4): 38–51.

Schmieder, Christian, Heiko Hesse, Benjamin Neudorfer, Claus Puhr, and Stefan W. Schmitz. 2012. “Next Generation System- wide Liquidity Stress Testing.” IMF Working Paper 12/3, In-ternational Monetary Fund, Washington, DC. https://www .imf.org/external/pubs/cat/longres.aspx?sk=25509.0.

———. 2018a. “Euro Area Policies: Financial Sector Assessment Program Technical Note— Supervision and Oversight of Cen-tral Counterparties and Central Securities Depositories.” IMF Country Report 18/227, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/CR/Issues /2018/07/19/ Euro- Area- Policies- Financial- Sector- Assessment- Program-Technical-Note-Supervision-and-46101.

———. 2018b. “Euro Area Policies: Financial Sector Assessment Program Technical Note— Stress Testing the Banking Sector.” IMF Country Report 18/228, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2018/07/19/ Euro- Area- Policies- Financial- Sector -Assessment- Program- Technical-Note-Stress-Testing-the-46102.

———. 2018c. “Peru: Financial System Stability Assessment.” Country Report 18/238, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/CR/Issues /2018/07/25/ Peru-Financial-System-Stability-Assessment -46119.

———. 2018d. “Brazil: Financial Sector Assessment Program Technical Note on Stress Testing and Systemic Risk Analysis.” IMF Country Report 18/344, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2018/11/30/ Brazil- Financia l- Sector- Assessment -Program-Technical-Note- on-Stress-Testing-and-Systemic -46416.

———. 2019. “Australia: Financial Sector Assessment Program, Technical Note— Stress Testing the Banking Sector and Sys-temic Risk Analysis.” IMF Country Report 19/51, Interna-tional Monetary Fund, Washington, DC. https://www.imf .org/en/Publicat ions/CR /Issues/2019/02/13/Austra l ia -Financia l-Sector-Assessment-Program-Technica l- Note -Stress-Testing-the-Banking-46608.




https://www.imf.org/external/pubs/cat/longres.aspx?sk=25509.0.

https://www.imf.org/external/pubs/cat/longres.aspx?sk=25509.0.









https://www.imf.org/en/Publications/CR/Issues/2019/02/13/Australia-Financial-Sector-Assessment-Program-Technical-Note-Stress-Testing-the-Banking-46608










https://www.imf.org/en/Publications/WP/Issues/2019/02/21/Growth-at-Risk-Concept-and-Application-in-IMF-Country-Surveillance-46567




PART I

Principles


CHAPTER 2

Macro- Financial Stress Testing: Principles and Practices

HIROKO OURA • LILIANA SCHUMACHER

WITH

JORGE CHAN- LAU • DIMITRI G. DEMEKAS • DALE F. GRAY • HEIKO HESSE • ANDREAS A. JOBST • EMANUEL KOPP • SÒNIA MUÑOZ • LI LIAN ONG • CHRISTINE SAMPIC • CHRISTIAN SCHMIEDER • RODOLFO WEHRHAHN

The global financial crisis drew unprecedented attention to the role of stress testing of financial institutions in macroprudential and micropru-dential surveillance, and its role as an integral element of crisis management to inform policies aimed at restoring confidence in the financial

system. Current stress testing practices, however, are not based on a systematic and comprehensive set of principles but have emerged from a trial- and- error approach and practical expediency. The chapter draws on the experience gained from a decade of stress testing in the context of IMF Financial Sector Assessment Programs to propose seven “best practice” principles that are universally applicable, including (1) the intended scope of the stress testing exercise affecting the identification of risks and measurement of vulnerabilities, (2) the macro- financial channels through which shocks are transmitted, (3) the availability of risk- mitigating features, and (4) the effectiveness of communicating the findings. These princi-ples serve as practical guidance on how to tailor stress tests to specific circumstances, including the degree of financial sector development, busi-ness models, and the macroeconomic environment in which financial institutions operate.

This chapter is based on an IMF Policy Paper (IMF 2012b) prepared by Hiroko Oura and Liliana Schumacher, with contributions from Jorge Chan- Lau, Dimitri Demekas, Dale Gray, Heiko Hesse, Andreas Jobst, Emanuel Kopp, Sònia Muñoz, Li Lian Ong, Christine Sampic, Christian Schmieder, and Ro-dolfo Wehrhahn. The chapter draws on a survey of stress testing practices at central banks and national supervisory authorities designed by Li Lian Ong, Hiroko Oura, and Liliana Schumacher and summarized by Ryan Scuzzarella (IMF 2012c).1 While advanced techniques to identify risks have gained increasing prominence, the fundamental design and structural logic of stress tests (such as nature and

severity of stresses as well as the role and influence of expert judgments) are at least equally important, and should be the primary focus of the design of the stress test(s).

market participants. The experience highlighted the useful-ness of stress tests as a diagnostic tool, but also revealed weaknesses in stress tests undertaken prior to the crisis by the banks themselves, supervisory authorities, and the IMF, all of whom to a greater or lesser extent failed to capture the risks that eventually materialized. In particular, a key lesson from the crisis has been a greater focus on concepts to iden-tify the buildup of financial risks. This has spawned risk- based framework(s) for financial stability analysis, including the examination of macro- financial linkages and the inte-gration of advanced market and risk- based tools for surveil-lance purposes.1 At the same time, the crisis underscored the potential of credible and comprehensive stress tests in restor-

1. STRESS TESTING: A PRIMER Over the last 20 years, stress testing has become essential to financial stability analysis. Stress tests first emerged in the late 1990s and have been used since then by financial insti-tutions, regulatory bodies, and international organizations such as the IMF and the World Bank, with the aim of proac-tively identifying vulnerabilities, and/or determining spe-cific risks for industry sectors, certain business models within these sectors, or systemically relevant institutions.

The global financial crisis placed a spotlight on stress test-ing of financial institutions, notably banks. The financial crisis had a significant impact on the way stress tests are be-ing carried out, not only by national authorities but also by


12 Macro- Financial Stress Testing: Principles and Practices

(Appendix 2.1). Greenlaw and others (2012) suggest princi-ples for stress tests focusing on risks that could have system- wide and economy- wide implications, but their principles are more conceptual, aimed at shifting the thinking about the purpose and goals of stress tests away from their micropru-dential focus on individual institutions toward systemic risk. These considerations are also echoed in the first Annual Report of the Office of Financial Research (OFR) at the US Treasury. Although this chapter concentrates mainly on stress tests con-ducted for macro- financial surveillance purposes— the key interest for the IMF, central banks, and macroprudential authorities— much of the discussion also applies to stress tests undertaken for other purposes, such as microprudential over-sight and institution- specific risk assessment.

In addition, this chapter provides the basis for a more sys-tematic approach to stress testing in FSAPs. The proposed principles establish a yardstick against which individual stress testing exercises can be evaluated, as well as an agenda for improvements to the IMF’s stress testing toolkit. These elements provided important input into the last review of the FSAP (IMF 2013a and 2014a).2

The rest of the chapter is organized as follows. Section 2 presents a brief introduction to the basic concepts and tools of stress testing. Section 3 discusses the lessons from the global financial crisis and European sovereign debt crisis for stress testers. Section 4 presents seven “best practice” princi-ples for stress testing and examines how closely actual stress testing practice corresponds to them. The principles are based not only on the IMF’s own extensive experience, but also that of its member countries, on the basis of a survey undertaken for this purpose.3 Finally, key conclusions and practical implications of these principles for stress testing practitioners are presented in the fifth section.

2. WHAT IS STRESS TESTING?Stress testing is a technique that measures the vulnerability of a portfolio, an institution, or an entire financial system under different hypothetical events or scenarios. It is a

ing market confidence in the financial system, as demon-strated by the Supervisory Capital Assessment Program (SCAP) exercise undertaken by the US authorities in 2009. Stress testing, once an arcane subject, has become almost a household name.

As a result of the attention and lessons learned from the crisis, the approaches, underlying assumptions, and uses of stress tests are being scrutinized and actively debated. The large and complex menu of choices in each of these areas has given rise to questions about the interpretation and consis-tency of findings generated by stress testing exercises. And the continued fragilities in financial systems during the transition to a more robust postcrisis regulatory framework have made the communication of stress test results an in-creasingly sensitive issue for both supervisors and financial institutions struggling to balance the need for greater trans-parency with the need to avoid alarming markets and creat-ing self- fulfilling prophecies.

The IMF is well- placed to contribute to this debate, hav-ing amassed significant practical experience in applying stress tests in a wide range of countries over a decade. Stress testing of financial systems has been a key component of the Financial Sector Assessment Program (FSAP) launched in 1999 and, subsequently, part of the analytical tools used in the biannual Global Financial Stability Report (GFSR). The IMF staff has also played an instrumental role in developing and disseminating several advanced stress testing models and cooperates closely with technical experts in supervisory agencies and central banks in testing and implementing new techniques, including through the Expert Forum on Ad-vanced Stress Testing Techniques and the Research Task Force (RTF) of the Bank for International Settlements (BIS). And the IMF is providing technical assistance and training to member countries interested in building or expand-ing stress testing capabilities.

This chapter discusses current practices in stress testing and proposes operational “best practice” principles for their design and implementation. These practical guidelines are derived from the years of the IMF’s own experience in devel-oping and using stress testing tools, including the regular internal review and evaluation of FSAP stress tests. They are not meant to be a “general theory” of stress testing or pro-vide a comprehensive step- by- step stress testing manual. In-stead, the key goal is to help set realistic expectations about the effective application of stress tests and explore how their design and implementation can be improved to ensure that they remain useful tools in identifying financial sector vulnerabilities.

The chapter fills a major gap in this debate. To the IMF staff’s knowledge, this is the first time that specific, opera-tional principles have been put forth for system- wide stress tests. The BIS and Committee of European Banking Super-visors (CEBS)—the predecessor of the European Banking Authority (EBA)—have also proposed principles for stress testing, but those were mainly directed to banks performing stress tests as part of their riskmanagement functions

2 In September 2010, the Executive Board of the IMF made financial stability assessments under the FSAP a regular and mandatory part of the bilateral surveillance under Article IV for jurisdictions with system-ically important financial sectors.

3 The survey was conducted in November 2011 and covered (1) the broad use and definition of stress tests; (2) banking sector stress tests (process and organization of stress tests; framework for solvency tests, including risk/scenario selections, macro- financial linkages, determining capital adequacy; framework for liquidity stress tests; communication strat-egy); and (3) issues regarding the use of stress tests and their application to the nonbank financial sector. A total of 26 central banks and super-visory authorities from 23 different countries responded (in some coun-try cases, the responses were jointly submitted by more than one agency). Among the 23 countries, seven were emerging market and de-veloping economies (EMDEs) and 16 were developed economies, 13 were European, six were Asian, and four were Western Hemisphere countries. A separate background paper (IMF 2012c) provides details of the survey results.


Hiroko Oura and Liliana Schumacher 13

uncertain and depend on economic and financial condi-tions. For an institution to be solvent as a going concern, it would need to maintain a minimum of positive equity capi-tal so that it can absorb potential losses in the event of a shock. Higher amounts of capital than this minimum might be needed to ensure continued access to market funding at a reasonable cost.

A solvency test assesses whether a firm has sufficient capi-tal to remain solvent in a hypothetically challenging envi-ronment by estimating profit, impairment losses, and valuation changes. The main risk factors are potential losses from borrowers’ default (credit risk) and securities due to changes in market prices such as interest rates, exchange rates, and equity prices (market risk). A stress test may ex-amine the impact of individual risks (single factor tests) or multiple sources of risks (multiple factor tests). Risk factors could be combined in an ad hoc manner (combined shock test) or generated more coherently using a macroeconomic framework (macroscenario tests).

Solvency tests may cover varying segments of a balance sheet (Figure 2.1). A test for credit risk may cover total loans (including interbank lending) or loans to certain segments (such as corporate, mortgages, or credit card loans). Market risk is assessed for securities in the held- for- trading (HfT) and in the available- for- sale (AfS) accounts, but loans, ad-vances, and debt securities in held- to- maturity (HtM) ac-count may be excluded, because these securities are supposed

quantitative “what if” exercise, estimating what would hap-pen to capital, profits, cash flows, of individual financial firms or the system as a whole if certain risks were to materialize.

A complete stress testing exercise is more than just a nu-merical calculation of the impact of possible shocks. It in-volves (1) choices on the coverage of institutions, risks, and scenarios; (2) the application of a quantitative framework to link various shock scenarios to solvency and liquidity mea-sures; (3) a strategy for the communication of the results; and (4) follow- up measures, if warranted. In this chapter, the term “stress test” refers to the entire process.

Stress tests typically evaluate two aspects of a financial institution’s performance: solvency and liquidity. As most stress tests so far focus on the banking sector, it will be the main focus in the rest of this chapter as well. Emerging stress testing techniques for nonbank sectors, notably insurance and financial market infrastructures, are never-theless becoming increasingly important, and are reviewed in Appendix 2.2.

Solvency Tests

An institution is solvent when the value of its assets is larger than its debt, that is, there is a certain amount of positive equity due to a positive net asset value. The values of both assets and liabilities depend on future cash flows, which are

Source: Authors.Note: SMEs = small and medium- sized enterprises.

Figure 2.1 Simplified Bank Balance Sheet

Assets Liabilities

Cash and cash equivalent Central bank loans

Moneymarketassets

Interbank lending Moneymarketliabilities

Interbank borrowing

Reverse repurchase agreements and securities borrowing

Repurchase agreements andsecurities lending

Certificates of deposit, commercial paper

Certificates of deposit, commercial paper

Securities Held-for-trading Customer deposits (financial institutions,public sector, firms/SMEs, and household)

Available-for-sale equities and debt Long-term borrowing

Held-to-maturity securities Debt instruments

Customer loans (�nancial institutions, public sector, corporate, and household) and related-party lending

Derivatives

Derivatives Other borrowings

Other assets Equity

Off-balance-sheet items • Derivatives • Contingent claims and liabilities (credit lines, guarantees, [implicit] guarantees to special purpose vehicles) • Securitization, resecuritization exposures


Macro- Financial Stress Testing: Principles and Practices14

key risk parameters (net operating profit, nonperforming loan ratio, probability of default,6 loss given default, and provisioning7) and relevant macroeconomic variables, such as GDP, unemployment, exchange rate, asset prices, and in-terest rates. It requires the use of econometric models— as well as considerable judgment— and the IMF provides tech-nical assistance in this area. Macroscenario stress tests cover several years (typically one to three in the case of country supervisory authorities or central banks, and often longer in FSAPs), as credit risks materialize gradually in economic downturns. Therefore, preimpairment profits (profits before loan and security losses) also need to be projected, since re-tained earnings in the test horizon would affect capital. This requires making assumptions about bank behavior (such as dividend payout policies and deleveraging, in case of ad-verse shocks), which introduces significant degrees of freedom— and complexity— to the exercise (Jobst, Ong, and Schmieder 2013).

Liquidity Tests

Liquidity stress tests aim to capture the risk of a bank failing to generate sufficient funding from cash inflows to satisfy short- term payment obligations arising from a sudden real-ization of liabilities in a stress scenario (Jobst, Ong, and Schmieder 2017). The tests assess the adequacy of the avail-able funding sources over a defined stress horizon.

Financial institutions may encounter sudden cash out-flows, for instance, because of:

• Sudden distress with their funding. Financial intermedi-aries that engage in liquidity/term structure transfor-mation, particularly banks, have, by the nature of their business, maturity mismatches in their balance sheets. Thus, they need to carefully manage scheduled and unscheduled cash outflows (including the loss of fund-ing sources) against cash inflows that are related to ma-turing assets, the rollover risk stemming from any existing maturity mismatches, as well as the ability to access unsecured retail/wholesale funding markets. If a large amount of deposits is suddenly withdrawn, or funding markets (such as repurchase agreements and commercial paper) freeze, a bank might no longer be able to meet its current and future cash flow needs even if it is otherwise solvent (“funding liquidity risk”). This would also involve market liquidity risk if the banks cannot sell assets quickly owing to deterioration in its liquidity (“market liquidity risk”).

to be paid in full at maturity as long as the debtor does not default.4 Off- balance- sheet exposures, including contingent claims and securitization exposures, could be affected by both market and credit risks and could potentially have highly nonlinear responses to stress. Bank solvency tests usually do not adjust the value of liabilities for changes in interest rates (market risk) due to their positive duration gap, that is, most bank liabilities are short- term deposits and money market instruments funding lending and investment over longer terms. On the other hand, stress testing for in-surance companies and pension funds considers such adjust-ments, as their liabilities have long maturities, and their present value depends on interest rates.

Solvency is measured by various capital ratios, typically following regulatory requirements. The capital impact of the solvency test is reflected in changes of the regulatory capital ratio (capital adequacy) and the extent to which it would drop below the prudential minimum under the assumed stress scenario, resulting in a capital shortfall. Standard choices for banks are the ratio of statutory/core Tier 1/Tier1 capital to risk- weighted assets (RWA); leverage ratios (capital to assets); losses in percent of capital; or capital shortfalls (the amount of capital needed to maintain a certain capital ratio). Individual institutions or the system at large are said to “pass” or “fail” the test if the target capital ratio is above a predetermined threshold or “hurdle rate.” Hurdle rates are often set at the current minimum regulatory requirement, but they could be set at different values if circumstances warrant, for example when new regulations (for example, Basel III) are expected to be introduced or to maintain a certain level of market funding cost.5 The choice of the hur-dle rate is a critical factor in stress testing exercises, especially when the results of the tests are directly linked to capital planning directed by supervisors.

Estimating solvency ratios in macroscenario stress tests requires estimating macro- financial models. A macro- financial model estimates the empirical relationship between

4 Values of securities in trading and AfS accounts are mostly assessed at market values ( mark- to- market [MtM] valuation). Losses and gains from trading securities are accounted for in the profit- and- loss state-ment (and therefore affect capital). Unrealized capital gains and losses from AfS securities also affect regulatory capital (that is, Core Tier 1 capital), albeit at varying degrees depending on national regulation and accounting rules. Unrealized losses should in principle be deducted from regulatory capital under the Basel III framework. However, the Basel II framework does not refer specifically to the treatment of unreal-ized AfS losses, granting national supervisors substantial national dis-cretion in applying prudential filters for gains and losses of AfS securities (“AfS filter”). Under the transitional arrangement of Basel III, a rising fraction of unrealized losses is deducted from capital (in recog-nition of existing provisions). A standard historical cost approach is ap-plied to HtM securities, which are valued at amortized cost net of any impairment provision (that is, net book value) unless there are persistent and substantial unrealized losses. The precise valuation practices differ across jurisdictions.

5 Pillar 2 of the Basel II framework empowers supervisors to request above minimum requirement capital ratios in line with banks’ risk profiles.

6 Point- in- time (PIT) or through- the- cycle (TTC) default probabilities may be used, but the latter tend to dampen portfolio risk (Rosch and Scheule 2008).

7 Under Basel I or the standardized approach of Basel II, the amount of loss provisions depends on the historical losses (nonperforming loan ra-tio) and loan classification. Under the internal ratings- based (IRB) or advanced IRB approaches of Basel II (and Basel III), however, banks need to determine provisions based on complete historically credit risk dynamics (that is, probability of default/loss given default or credit rat-ing data).



minimum capital regulation (Pillar 1). Additional tests can be required in the context of Pillar 2 that provides supervisors powers to order management actions by banks if deemed necessary. A survey by the Basel Committee for Banking Supervision (BCBS) (BCBS 2012) indicates that supervisory stress tests are increasingly utilized to set capital re-quirements for specific banks, determine explicit capital buffers, or limit capital distributions by banks in the context of the capital review process under Pil-lar 2 of the Basel framework. The liquidity risk stan-dards established by the Basel III (and the Solvency II regime for European insurance companies) utilize stress testing as an integral part of the regulatory framework.

• Macroprudential/surveillance stress testing. Over the past two decades, many country authorities have started using stress test exercises to assess system- wide risks, in addition to institution- specific risks, based on the capital adequacy of the banking sector under adverse macroeconomic conditions. The re-sults are often reported in their financial stability re-ports. As an integral part of risk- based analytical frameworks supporting macroprudential policy, sur-veillance stress testing aims at identifying systemic risk ex ante, thereby minimizing the incidence and impact of disruptions in the provision of key finan-cial services that can have adverse consequences for the real economy (and broader implications for eco-nomic growth).9 Such risk to financial stability arises from the collective impact of common shocks on systemically important institutions and/or a material number of firms, possibly amplified by market fail-ures and/or fault lines in the architecture of the fi-nancial system, for instance between banking and nonbanking financial sector activities. The IMF has regularly included macroprudential stress testing in FSAPs since the program’s inception in 1999 (Jobst, Ong, and Schmieder 2013). A few country authori-ties indicated in the staff survey (IMF 2012c) that the FSAP stress tests were the first such exercises un-dertaken in their countries.

• Crisis management stress testing. Stress tests have also been used, especially after the global financial crisis, to assess whether key financial institutions need to be recapitalized or not, possibly with public support. In particular, the SCAP exercise in the United States and the EU system- wide stress test organized by the CEBS/EBA in 2010 and 2011 attracted attention, because banks were required to recapitalize based on the test results, and the detailed methodology and individual banks’ results were published. In IMF programs with banking sector distress (including

• Interlinkages between market and funding liquidity risk. Financial institutions active in taking market positions may face a sudden liquidity need when asset markets become volatile, increasing initial/variation margin requirements. For example, those trading in highly leveraged derivatives markets may face a li-quidity shortage if a position becomes out- of- the- money8 (even if at expiration this result is reversed) or gets downgraded. Even when these positions are taken by legally separate special purpose vehicles (SPVs), some institutions may be forced to support these SPVs for reputational reasons, effectively inter-nalizing the liquidity shortage.

Financial institutions encounter a liquidity shortage when they cannot generate sufficient cash in response to a shock. Banks that have enough liquid assets counterbalanc-ing capacity can generate sufficient cash either by selling them or by pledging them as collateral for repurchase agree-ments without making large losses. However, if available as-sets are mostly nonmarketable loans, or if the market value of these assets declines substantially below the book value (haircut), banks may face liquidity constraints. Several possible hurdle rates may be used in a liquidity stress test, such as the number of days the institution can tolerate a li-quidity shock before net cash flows become negative (which would also manifest in a deterioration of stressed liquidity ratios).

Liquidity and solvency stress events are often closely in-terrelated and hard to disentangle. In the event of funding distress, a liquidity shortage may threaten the bank’s sol-vency if assets cannot be sold or can be sold only at loss- making prices (“fire sale”). Thus, higher market funding risk in a liquidity stress event is a factor that could translate into solvency stress.

The Typology of Stress Tests

There are four types of stress tests, which are differentiated by their ultimate objectives (for details, see Table 2.1):

• Stress testing as an internal riskmanagement tool. Fi-nancial institutions use stress testing to measure and manage the risks with their investments. One of the early adopters was J.P. Morgan in the mid- 1990s, which used value at risk (VaR) to measure market risk. However, early stress testing had limited cover-age of risk factors and exposures and little integra-tion with the overall risk management and business and capital planning at firms.

• Microprudential/supervisory stress testing. The Basel II framework requires banks to conduct stress tests for market risk and, in some cases, credit risk as a part of

8 The term “ out- of- the- money” refers to a call option with a strike price that is higher than the market price of the underlying asset, or a put option with a strike price that is lower than the market price of the un-derlying asset. 9 See also Borio and Drehmann 2009.


Macro- Financial Stress Testing: Principles and Practices

16

TABLE 2.1

Typology of Stress TestsFeatures Macroprudential (Surveillance) Microprudential (Supervisory) Crisis Management Internal Risk Management

Main objective Unveil the sources of systemic risk and vulnerability in the context of surveillance and regular system-wide monitoring

Assess the health of an individual institution; inform supervision of the institution

Input for bank recapitalization and business restructuring plans

Manage risks from existing portfolio; input for business planning

Organized by Central banks, macroprudential authorities, IMF

Supervisors (microprudential authorities)

Macro and microprudential authorities

Financial institutions

Coverage of institutions All (or as many as possible) institutions, especially systemi-cally important institutions

Supervised individual institutions (tests for different banks could take place at different times)

Varies, but it should include all distressed and near-distressed institutions.

Own activities

Frequency Typically annual or semiannual for country authorities, or in the context of FSAPs

Individual institutions are tested as needed; increasing number of supervisors conduct regular stress tests (with common assumptions)

As needed High (daily or weekly) for market risks; lower for enterprise-wide exercises

Nature of shocks Systemic and common shocks across institutions; shocks tend to be extreme

Often idiosyncratic; common macro assumptions are sometimes made for horizontal or thematic review across institutions

Ongoing systemic stress (baseline) or relatively mild shocks, mainly focusing on solvency risks

Idiosyncratic or systemic (those that matter for the particular institution)

Capacity to incorporate systemic risks

Through macro and market-level shocks and additional system-wide features (for example, network effects)

Through macro and market-level shocks



Likelihood of assumed shocks Low Low High VariesAssessment criteria (hurdle rates) Current or prospective regulatory

requirements or alternative thresholds (if appropriate)

Current or prospective regulatory requirements or alternative thresholds (if appropriate)

Current or prospective regulatory requirements or alternative thresholds (if appropriate)

Internal risk tolerance indicators and regulatory requirements

Key output metric Aggregate indicators for the system and their dispersion

Individual institution indicators Individual institution indicators Individual institution indicators

Follow-up measures after tests Typically no follow up for individual institutions, but often used as the basis for discussion of potential macroprudential or system-wide measures

Institutions with weak results are often required to explain and take management actions if deemed necessary by supervisors

“Failing” institutions are often required to take major manage-ment action, such as recapitaliza-tion, possibly with government support

May or may not require management action

Publication Often Rarely Varies NoExamples FSAPs, GFSRs, other financial stability

reports Dodd-Frank Act Stress Test as part of

CCAR (United States), tests required by Basel framework, ECB-Banking Supervision stress test (SSM)

SCAP (United States), EU system-wide stress test (CEBS/EBA), exercises in some IMF program countries

RiskMetrics (J.P. Morgan’s VaR model)

Source: Authors.Note: CCAR = Comprehensive Capital Analysis and Review; CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; ECB = European Central Bank; SSM = Single Supervisory Mechanism; FSAP = Financial Sector Assessment Program; GFSR = Global Financial Stability Report; SCAP = Supervisory Capital Assessment Program; VaR = value at risk.



Stress testing in FSAPs and by the national authorities may be conducted either as top- down or bottom- up exercises or both. Top- down exercises are defined as those conducted by the national authorities or IMF staff (typically as part of FSAPs) using bank- by- bank data and applying a consistent methodology and assumptions. Bottom- up exercises are car-ried out by individual financial institutions (with or without involvement of supervisory authorities) using their own internal data and models, often under common assump-tions. Some supervisory tests include bank- specific risks11 (Table 2.2), including reverse stress tests based on shocks that could render a specific institution insolvent. Liquidity tests are often conducted as bottom- up exercises because they require granular data and depend on banks’ liquidity strategies, and banks are given more flexibility regarding de-tailed assumptions compared to solvency tests. FSAPs al-most always include a top- down test, frequently supplemented by a bottom- up test. Many national authorities use both top- down and bottom- up tests and emphasize the importance of running their own top- down tests to effectively validate bottom- up results.

Communication practices differ across the four types of stress testing. Close communication between banks and su-pervisors, between supervisory agencies within a country, or between FSAP teams and country authorities is required for effective stress tests. Public communication of stress test

Ireland, Greece, and Portugal), estimating bank recapitalization needs through stress tests was an im-portant component. As this use of stress tests as a crisis management tool is relatively new, Appendix 2.3 presents the key features, similarities with, and differences from other types of stress tests, using three prominent examples of this type of stress tests (US SCAP, the EU system- wide stress test in 2010 and 2011, and the EBA Capital Assessment Exercise of 2012) as illustrations.10

The risk coverage and methodologies have evolved over time, as the use of stress tests is broadened. Financial institu-tions are now expected to manage enterprise- wide risks, which cover a broad range of exposures and risk factors in an integrated manner, crossing over the internal segmentations of various business lines. Similarly, macroprudential stress tests have evolved from single- factor tests to macroscenario tests.

Depending on the objective of stress testing, follow- up managerial or supervisory actions may be taken. Macropru-dential tests, including in FSAPs, typically do not prescribe bank- specific action, although they could lead to macropru-dential policy recommendations. Supervisory stress tests are increasingly used to guide supervisory action, ranging from improving data collection, targeting examinations, and closer monitoring, to requiring bank management actions, such as raising additional capital, reducing certain expo-sures, capping dividends, and updating individual institu-tions’ resolution plans. Follow- up is almost certain in crisis management stress tests, which are expressly designed to es-timate capital shortfalls.

11 There is some variation in bottom-up practices among country authori-ties, including bottom-up tests being conducted with institution- specific assumptions or implemented by the supervisory authority using bank-by-bank data.

12 In the United States, the Dodd- Frank Act requires communication of Comprehensive Capital Analysis and Review results by both the US Federal Reserve Board and individual banks (Board of Governors of the US Federal Reserve System 2009 and 2012).10 See also Ong and Pazarbasioglu 2014.

TABLE 2.2

Comparison of Bottom-Up and Top-Down Stress Tests

Feature Bottom Up Top Down

Strengths • Reflects granular data and covers exposures and risk- mitigating tools more comprehensively, including those that are hard to cover in top-down tests (such as risks from complex structured products, hedging strategies, and counterparty risks).

• Utilizes advanced internal models of financial institutions, which could potentially yield better results.

• May reveal risks that could otherwise be missed.• Provides insights in the riskmanagement capacity and

culture of a particular institution.• Application of severe common shocks may encourage

individual institutions to prepare for tail events that they might otherwise not be prepared to contemplate.

• Ensures uniformity in methodology and consistency of assumptions across institutions.

• Ensures full understanding of the details and limitations of the model used.

• Provides an effective tool for the supervisory authority or IMF Financial Sector Assessment Program team to validate bottom-up tests.

• Once a core framework is in place, implementation is relatively resource-effective.

• Can be implemented in systems where institutions have limited risk management capacities.

Weaknesses • Implementation tends to be resource intensive and depends on the cooperation of individual institutions.

• Results may be influenced by institution-specific assump-tions, data, and models that hamper comparisons across institutions.

• Estimates might not be precise due to data limitations. • Standardization may come at the cost of not reflecting

each institution’s strategic and managerial decisions.

Source: Authors.



and Table 2.3 presents a detailed comparison of their opera-tional aspects.13

Model choices are somewhat constrained in EMDEs with limited data or weak accounting and regulatory systems, but some options are nonetheless available. At the same time, these countries tend to have simpler financial and economic systems, and key economic risks and vulnerabilities are rela-tively straightforward to identify. Simple balance- sheet- based tests (such as those illustrated in Čihák 2007) with single or multifactor shocks can be implemented in most countries us-ing basic supervisory data. In cases where supervisory data are patchy or unreliable, or the magnitude of uncertainty is too large to draw strong conclusions from the tests, the prior-ity should be to improve supervision and data quality rather than develop stress testing approaches. Developing a macro- financial linkage model for macroscenario tests can also be a challenge in some EMDEs, given limitations in the quality or availability of long time series data. One option in such cases is to utilize cross- country data.14

3. LESSONS FROM THE GLOBAL FINANCIAL CRISIS AND THE EUROPEAN SOVEREIGN DEBT CRISIS Both the global financial crisis and the European sovereign debt crisis had major impacts on the design, completion, and interpretation of stress tests. First, the global financial crisis revealed some weaknesses of the preexisting stress testing approaches. There is broad agreement that most stress testing exercises before the crisis— whether conducted by industry, national authorities, or the IMF staff in the context of FSAPs— failed to detect key macro- financial vul-nerabilities (Haldane 2009a, 2009b). Second, the crises spurred the use of stress tests for crisis management pur-poses, notably in the United States and Europe, albeit with mixed results (Appendix 2.3). Third, these stress tests en-gendered a move toward routine system- wide stress testing for surveillance purposes, with stress test results being shared more widely.

So why did precrisis stress tests miss the vulnerabilities that eventually materialized?

results, on the other hand, is not common, although this has been changing, especially for crisis management stress tests.12 Macroprudential/surveillance stress test results are typically reported in financial stability reports— or Finan-cial System Stability Assessments (FSSAs) in the case of FSAPs— but in varying degrees of aggregation, usually with-out identifying individual institutions. In general, dissemi-nation of stress test results is controversial. Several country authorities have voiced concerns that public dissemination might create unrealistic expectations, lead to misinterpreta-tions in the mass media, and potentially detract from the value of stress tests as a supervisory tool, as banks focus too much on the media impact. For FSAPs, the publication of FSSAs is voluntary, but the majority of countries— including most of the jurisdictions with systemically important finan-cial sectors— publish the documents. The accompanying stress testing Technical Notes, which are more detailed, are published much less frequently.

Stress Testing Models

Stress tests use a wide range of analytical models. These models are designed to capture key stress factors to solvency or liquidity of individual institutions or financial systems, varying widely in terms of complexity and data require-ments. They can be broadly classified into two categories: (1) models predicated on a detailed analysis of balance sheets of individual institutions (sometimes called “fundamental” ap-proaches), and (2) models based on summary default mea-sures for individual portfolios, institutions, or entire systems embedded in market prices, such as stocks, bonds, and derivatives.

Both approaches are complements rather than substi-tutes. Balance- sheet- based approaches can identify the source of individual vulnerabilities in relation to the ac-counting identities provided in prudential and/or statutory reporting. Thus, they are more detailed and informative as to the effectiveness of potential remedial actions flowing from the findings of stress tests, and can be applied to emerging market and developing economies (EMDEs), where stock markets tend to be thin or illiquid (and only few financial institutions are publicly listed). But they are backward- looking, data- intensive, hard to update fre-quently, and not particularly suited for capturing interde-pendence (portfolio) and contagion effects across institutions. In contrast, market- price- based approaches are more flexible, can easily incorporate portfolio effects and risk factors as perceived by the market, and can be updated as frequently as desired. However, they also involve an iden-tification challenge, which, depending on the model choice, can make it difficult to determine the precise source of vul-nerabilities; in addition, market- based risk metrics are sensi-tive to short- term swings in market perceptions that may have little to do with fundamentals, and cannot be applied to countries or entities with limited or no market price data. Box 2.1 presents an overview of the two families of models,

13 The development of stress testing models has gathered momentum in the wake of the global financial crisis, supporting a plurality of different approaches with their own strengths and weaknesses. As the discussion in some of the following sections illustrates, the IMF staff is very active in this area, in close cooperation with advanced economy central banks and financial stability agencies and the academia. The annual Expert Forum on Advanced Stress Testing Techniques, co- organized by the IMF’s Monetary and Capital Markets Department, is one of the pre-eminent fora for exploring some of these new approaches among stress testing practitioners.

14 For instance, Annex 1.6 in the April 2010 GFSR (IMF 2010b) shows that a nonperforming loan projection model using data from Latin America and emerging market countries in Asia performs reasonably well in predicting loan quality in central and eastern European coun-tries, where time series are relatively short.



Box 2.1. Balance-Sheet-Based and Market-Price-Based Approaches

There are two general approaches to estimating the impact of macro-financial shocks on bank solvency: balance-sheet-based and market-price-based. Balance-sheet-based models cover all material on- and off-balance-sheet exposures and the risks to which these are exposed. In contrast, market-price-based models are based on summary bank default measures embedded in asset prices (such as bank stocks, bonds, and derivatives). These measures are extracted from market prices by solving for the default probability and asset recovery implicit in them, using standard pricing formulas. The IMF staff often uses the contingent claims approach, a market-price-based approach built on option pricing theory on a bank-by-bank basis and/or system-wide to cover joint default risk within a banking sector (Gray, Merton, and Bodie 2008). When used for stress tests, market-price-based methodologies need to project the market-based default measures for the period covered by the test.

Both approaches have advantages and disadvantages. Balance-sheet-based models are more informative but are backward-looking, data intensive, and hard to update given the typical time lag of accessible prudential/statutory information. In contrast, market-price-based measures are forward-looking, can be easily updated, and, in many cases, already incorporate portfolio effects (for example, the parametric approach of Credit Risk+ (Avesani and others 2008) as well as the Consistent Information Multivariate Density Optimizing cop-ula function (Segoviano 2006) and systemic contingent claims approach (Jobst and Gray 2013). This alternative valuation approach is fun-damentally driven by the intent to quantify possible linkages between financial assets and institutions in support of a more comprehensive assessment of common vulnerabilities than would be allowed by any single approach. However, they also reflect the impact of market swings, which may be unrelated to changing fundamentals.

While data limitations may dictate the preferred modeling approach in a specific set of circumstances, these two methodologies do not convey the same type of information and should be considered complementary rather than substitutes. Balance-sheet-based models are typically used by financial stability/supervisory authorities, who require disaggregated information on the sources of vulnerabilities to adopt risk-mitigating measures. Market-price-based approaches can be used when the focus is placed squarely on understanding the market assessment of banks’ loss absorption capacity under stress (which can also help supervisors prevent bias toward inaction for timely interventions that would not be possible using backward-looking information without any measure of uncertainty).

Because of their detailed nature, balance-sheet-based models face severe limitations in capturing all risks in an integrated manner. For this reason, in practice, some forms of risk have received overwhelming attention (such as default risk of private counterparties in the bank-ing book) at the expense of others (such as sovereign default risk; downgrading risk of sovereign and private counterparties; counterparty risks of derivatives; and liquidity risks, including funding costs). Methodologies aimed at the valuation of all positions held by banks and the assessment of all risks have been introduced (see, for instance, Barnhill, Papapanagiotou, and Schumacher 2002; Barnhill and Schum-acher 2011) but are data-intensive and require refinement. In addition, joint default risk within a system of firms varies over time and de-pends on the individual firm’s likelihood to cause and/or propagate shocks arising from the adverse change in one or more risk factors, which requires a closer examination of their historical volatility. Factors may be weakly correlated under normal economic circumstances but highly correlated in times of distress (as correlations become very high in downturn conditions).

A common challenge for both types of models is finding a way to stress test individual institutions to system-wide risks. Capturing the intrinsic dependencies across different types of risk is a major challenge for balance-sheet-based models. For example, counterparty risk is inherently dependent on the evolution of market risk; and the global crisis has shown that systemic liquidity risk cannot be assessed with-out considering the solvency profile of institutions. If risk factors are not fully correlated, it is reasonable to account for their dependence structure and combinations of stress testing parameters in which the individual impact of each risk is lower than the appropriate percentile for that risk in isolation (that is, the magnitude of change of a single risk factor relative to historical experience). This requires assessing measures of default dependence to produce estimates of systemic loss distribution. In the context of the balance-sheet-based models, this extension can be undertaken by superimposing a network of claims to keep track of default effects of one institution onto the others. Market-price-based models, on the other hand, typically treat the banking system as a portfolio of banks and derive a distribution of sys-temic losses using portfolio analysis techniques like those used for individual bank portfolios.

• The institutional perimeter of stress tests was too nar-row. Stress tests did not, as a rule, cover what came to be known as “shadow banking” (for example, money market funds and insurance companies writing credit insurance), which played a key role in originating or transmitting shocks.

• Key shock transmission and propagation channels were not covered. Interconnectedness among key financial institutions through cross- exposures propagated and amplified shocks (as in the case of Lehman Brothers). Second- round feedback effects between the financial sector, the real economy, and sovereign risk were not incorporated in macro- financial stress testing.

• Several important risk factors were not included. Shocks hit multiple markets and countries at the same time

due to common stress or contagion from one to an-other, which increased systemic risks. Several specific risk factors were also missed, including counterparty, basis, and contingent risks (BCBS 2009, 2012).15

• Balance sheet valuations did not fully reflect economic value. Stress tests based on regulatory and accounting norms overestimated the resilience of the financial

15 Counterparty risk is a type of credit risk (for example, risk of default of a counterparty to a derivatives transaction). Basis risk for hedging is the risk that hedging becomes imperfect because there is either (1) differ-ence between the asset, whose price is to be hedged, and the asset under-lying the derivative; or (2) a mismatch between the expiration date of the hedge and the actual selling date of the asset. Contingent risks could arise either from legally binding credit and liquidity lines or from reputational concerns related to, for example, off- balance- sheet vehicles.



system. Market pressures on sovereign and financial sector exposures, for instance, were not fully reflected in the balance sheet of banks, which did not value every security at market value. And hurdle rates in solvency stress tests were based on regulatory minima, which proved insufficient when markets demanded larger capital buffers due to higher uncertainty about bank valuations, resulting in some banks “passing” stress tests only to face severe distress shortly thereafter.

• The shocks were not severe enough and did not examine genuine “tail events.” In some cases, the stress scenar-ios turned out to be too benign compared to the ac-tual shocks. In other cases (for example, the unraveling of complex structured products), the em-pirical basis was too limited for calibrating the sensitiv-ity of these instruments to changes in macro- financial conditions affecting their credit performance. Thus, time- varying correlation and extreme market risks,

TABLE 2.3

Comparing Balance-Sheet-Based and Market-Price-Based ApproachesBalance-Sheet-Based Approach Market-Price-Based Approach

Primary input data • Accounting data (balance sheet, income statement, overview of interbank exposures)

• Financial market data (equity prices, bond yields, CDS spreads or equity option-based probability of distress)

Secondary input data

• Probability of distress (and loss- given-default) or nonperforming loan ratios/loan classification of borrowers (for credit risk)

• Market data (equity prices, exchange rates, interest rates, price volatility) to calibrate shocks

• Balance sheet data (combination of equity prices and accounting data to obtain key input variable, such as the Expected Default Frequency by Moody’s CreditEdge)

Type of test • Solvency, liquidity, and network analyses • Largely focused on solvency and its interdependence among key financial institutions; incipient research on testing for liquidity indicators (Jobst 2014) or liquidity stress

Frequency • Varies depending on the reporting cycle (quarterly, semiannual, annual)

• Daily or lower frequency

Application • Most banks or financial systems (including emerging markets and developing countries) as long as financial reporting or supervisory data exist and are available

• Limited to market data-rich countries and publicly listed institutions that is, cannot cover savings/cooperative banks, privately-held, or government-owned companies)*

• Stand-alone analysis for subsidiaries may be difficultLink(s) to

macroscenarios• Possible, by estimating additional macro- financial

model(s), linking macroscenario variables and risk factors (for example, probability of default of borrowers, nonperforming loan ratios)

• Possible, by estimating additional macro-financial model(s), linking macroscenario variables and risk factors (for example, probability of default of banks, volatility or leverage of banks)

Estimation of systemic effects

• By considering common macroshocks across banks (for example, GDP, inflation, and unemployment); and incorporating network effects (interbank exposures)

• By considering common macroshocks across banks and incorporating interdependence (portfolio) effects among banks, which may be estimated using asset prices

Output • Various capital ratios• Liquidity ratios• Capital shortfalls

• Expected losses• Unexpected losses• Probabilities of spillover among banks

Strength • Pinpoints the type of risk exposing the vulnerability (for example, credit losses from housing loans, market valuation losses from exposures to sovereigns, losses from currency mismatches)

• Possible to adjust for supervisory weakness (for example, under-provisioning, and forbearance)

• Less data intensive than the accounting-based approach• Focuses on systemic risks/losses and tail events• Incorporates risk factors priced by the market

Weakness • Data intensive (especially for network analysis)• Quality of the analysis depends on the granularity and

availability of the data.

• The causes of different risks are difficult to disentangle (“black box”)

• Estimated vulnerability measures may be very volatile during periods when markets are under significant stress, and links with balance sheet fundamentals may be obscured

Selected Examples

• Stress tester 101 (Čihák 2007)• Next generation stress testing (Schmieder, Puhr, and

Hasan 2011)• Network analysis (Espinosa-Vega and Solé 2010)• CreditRisk+ (Credit Suisse Financial Products)

• Systemic CCA (Jobst and Gray 2013)• CoVaR (Adrian and Brunnermeier 2008)• Distress dependence (Segoviano and Goodhart 2009)• SES and MES (Acharya and others 2010)

Source: Authors.Note: CCA = contingent claims analysis; CoVaR = conditional value at risk; MES = marginal expected shortfall; SES = systemic expected shortfall.* The IMF staff has developed an adjustment approach for nonlisted firms for the systemic CCA-based stress test in the context of the Financial Sector Assessment Program for Hong Kong, SAR (IMF 2014b).



This chapter proposes seven “best practice” principles for stress testing to address this issue with a view to promoting greater consistency and completeness. These principles are mainly focused on stress tests for macroprudential surveil-lance but they also contain elements that are generally ap-plicable to all types of stress tests. As such, they are related to, but do not overlap with, the Principles for Sound Stress Testing Practices (BCBS 2018), which focus on banks’ own stress testing practices (Appendix 2.1).16 The remaining sec-tions discuss the implications of these principles and evalu-ate to what extent actual practices correspond to them, based on the survey results (IMF 2012c). The seven principles are:

• Coverage. Define appropriately the institutional pe-rimeter for the tests.

• Risk transmission. Identify all relevant channels of risk propagation.

• Scope. Include all material risks and buffers based on the total balance sheet.

• Interpretation. Make use of the investors’ viewpoint in the design of stress tests.

• Calibration. Focus on tail risks.• Communication. When communicating stress test

results, speak smarter, not just louder.• Limitations. Beware of the “black swan.”

Principle 1 (Coverage): Define Appropriately the Institutional Perimeter for the Tests

This principle targets the selection of the institutions to be included in the tests. For system- wide tests, this involves a choice of which institutions to include and which to leave out. This choice requires an assessment of which banks are systemically important (that is, capable of triggering or am-plifying systemic risk). Size, substitutability, complexity, and interconnectedness are the criteria that are used to assess the systemic importance of internationally active banks (see IMF/BIS/FSB 2009; BCBS 2011). These criteria are often mirrored in the assessment of systemic importance in the domestic context. A bank’s distress or failure is more likely to cause damage to other banks, markets, or the economy if its activities comprise a large share of financial intermedia-tion. The larger the network of a bank’s contractual obliga-tions in which it operates, the higher the likelihood that its failure will materially impact the default probability at other institutions. The systemic impact of a bank’s distress or fail-ure is expected to be negatively related to its degree of substi-tutability as a market participant or service provider (in the

which were the main contributors to tail risks, were not well reflected in stress tests (Rosch and Scheule 2008).

• Many shocks to identified risk factors were considered “unthinkable,” and, thus, were not included in the stress test. For instance, severe liquidity risks, involving the closure of key funding markets and its impact on bank solvency, were not included in stress testing exer-cises. Similarly, sovereign default risk in advanced economies was not covered. Both scenarios were con-sidered too extreme to be plausible.

The design and implementation of stress tests should ide-ally be based on “best practice” principles that incorporate the lessons from past crises, include severe but plausible sce-narios, and be sufficiently operational. Such principles should provide practitioners with operational guidance on how to tailor stress tests to a wide variety of country and sector circumstances while maintaining minimum stan-dards that:

• ensure comparability across different exercises (which is particularly important for FSAPs (Jobst, Ong, and Schmieder 2013, 2017);

• protect against their most obvious pitfalls, especially considering the lessons from the global financial cri-sis; and

• interpret their results appropriately. Efforts are being made to improve stress testing frame-

works in light of this experience, but this is still an unfin-ished agenda. While it is important not to “fight the war of the past,” financial sector crises offer important insights into market dynamics (including the reaction function of key market participants) under stress. Current practices have in-corporated some lessons learned from past crises, notably in terms of the types of scenarios and the severity of shocks, partly because shocks that materialized during the global fi-nancial crisis provide a good benchmark for tail risks. How-ever, challenges remain with incorporating all relevant risk transmission channels, as well as feedback effects between the financial sector and the real economy. Appendix 2.4 presents some of the emerging methodological approaches at the frontier of stress testing techniques that attempt to tackle these challenges.

4. “BEST PRACTICE” PRINCIPLES AND ACTUAL STRESS TESTING PRACTICESStress testing practices have thus remained largely unsystem-atic. In many instances, established practices in various in-stitutions, including the IMF, often reflect constraints in human, technical, and data capabilities on the institutions undertaking stress testing. They have emerged mostly out of trial- and- error efforts, or, in some cases, even as a matter of habit or convenience. Existing approaches do not always re-flect a systematic effort to build a structured approach to stress testing.

16 The Basel Committee published these principles following public con-sultation (BCBS 2018). The new principles now apply to both supervi-sors and banks and are presented at a higher level than the previous ones to enhance broader applicability and preserve flexibility to accommo-date developments in stress testing practices. In addition, previous over-laps between principles were removed, and separate considerations for banks and supervisory authorities have been provided within each principle.



insurance companies are also becoming more common. Al-though traditional insurance activities are not very likely to trigger systemic risk,18 insurance companies may engage in nontraditional, noninsurance (NTNI) activities, such as nonhedging- related derivatives trading and securities lend-ing/repo, which can create systemic risk through their links with banks. In this regard, a network approach integrating all financial institutions would be less likely to leave out sources of potential systemic risk.

Applying this principle requires a good understanding of the system’s main features before undertaking the stress tests. This includes knowledge about the relevant market participants as well as their operations, business models, types of transactions, areas of risk concentration, and the likely channels of risk transmission. A formal mapping of this understanding into a network of claims and potential claims would facilitate the job, but other, more heuristic (and qualitative) approaches may also be used to get an un-derstanding of the system, including market intelligence and conversations with market participants.

How well do actual stress testing practices correspond to this principle?

• Bank stress tests tend to be comprehensive and cover ei-ther all institutions in the system or focus on systemically important institutions. FSAP stress tests typically cover at least 70 percent of the locally incorporated commercial banks (including subsidiaries of foreign banks and state- owned banks) by assets (Jobst, Ong, and Schmieder 2013). Country authorities’ exercises focus on private domestically owned commercial banks, followed by foreign subsidiaries and state- owned banks. Based on the IMF staff survey results (IMF 2012c), the coverage ranges from 60 to 100 percent of the system by assets (with the median being 85 percent and 16 banks). The number of banks in the sample varies from below five to over 1,000.

• However, the methodology used to define the perimeter of the most relevant institutions tends to vary. The size of the balance sheet and interconnectedness are key factors, followed by the size of local retail activities and legislative definitions. Formal network ap-proaches are increasingly used to assess interconnect-edness, although the more sophisticated models are still used in a minority of cases, even in advanced economies. Various indicators of interconnected-ness, in addition to size, played critical roles for de-termining the 25 jurisdictions with systemically

case where specific institutions provide critical services, such as a market infrastructure). Thus, the systemic relevance is higher for more complex institutions (all else being equal), as the cost and time needed to resolve them are greater.

While size, degree of substitutability, and complexity are observable features, an assessment of interconnectedness re-quires the use of sophisticated network approaches. Network approaches can be useful in identifying systemic institutions that should be covered by stress tests (Box 2.2). Financial in-stitutions hold claims against each other, which forms a net-work that can be thought of as a matrix of bilateral claims. Network models provide rich dimensions of interconnected-ness and can identify systemically important institutions or groups of financial institutions that are at the center of the network, going beyond simple metrics of cross- exposures. They can also be used directly to measure the likelihood of multiple defaults by systemically important institutions, a key feature of systemic risk (stress testing individual institu-tions’ solvency and simply aggregating the outcome would tend to underestimate systemic risk— see Principle 5).

Separately, stress testing requires identifying the appro-priate perimeter of financial activities that fall within the scope of the exercise. Some of the largest financial institu-tions have a wide range of activities, often straddling na-tional boundaries and covering different industries (banking, insurance, pension fund, investment fund, various financial SPVs, and even nonfinancial activities). The ultimate owner-ship structure and economic links may not be always clear. And although some of the SPVs may be legally independent, the crisis illustrated that they could still imply contingent liabilities for the “parent” companies if the latter choose to provide support beyond legal obligations for reputational reasons. Analyses of global banks in their home countries often tend to focus on their activities in that country, but cross- border exposures may be relevant in assessing inward and outward spillover effects. However, in the context of host countries, limited prudential information on group- wide activities available to host country supervisors might put the most important risk— the health of the parent company— outside of the scope of stress tests. The stress tes-ter needs to use judgment as to whether these activities should be consolidated or segregated.

Finally, defining the proper perimeter also calls for the inclusion of nonbank institutions that may trigger or propa-gate systemic risk, such as financial market infrastructures (FMIs) and insurance companies. The concern about FMIs, such as payment systems, central securities depositories, se-curities settlement systems, and central counterparties, is not their balance sheet per se, but rather their safe and reli-able functioning. Their systemic importance is given by the key role played by the services they provide.17 Stress tests of

17 For this reason, the Committee on Payment and Settlement Systems (CPSS) and the Technical Committee of the International Organiza-tion of Securities Commissions (IOSCO) requires an assessment of FMIs using stress testing.

18 Although traditional insurance activities have not contributed to sys-temic risk during the financial crisis, the IAIS (2013) identified some vulnerabilities from NTNI activities in its initial assessment methodol-ogy for the identification of globally active, systemically important in-surance firms. The weighted indicator- based approach for globally active, systemically important insurance firms is similar in concept to that used to identify global systemically important banks, but also in-troduces additional indicators that are germane to insurance activities.



Box 2.2. Network Models in Financial Stability Analysis and Stress Testing

Network models help measure interconnectedness, that is, linkages among financial institutions, systems, or entire countries, through claims held against each other or other channels. The importance of interconnectedness as a channel of contagion has been studied for some time. For example, Allen and Gale (2000), among others, explore how linkages among banks through direct exposures could be a source of contagion, which was underscored by the failure of Lehman Brothers in 2008.

Stress testing for financial market infrastructures also makes extensive use of network analysis. Nevertheless, it is important to keep in mind that the relationship between interconnectedness and financial stability is not simple and monotonic: interconnectedness may en-hance or reduce financial stability, depending on the degree of cross-institution or cross-border integration and the precise pattern of cross exposures or other linkages (Čihák, Muñoz, and Scuzzarella 2011).

Simple network models measuring interconnectedness through cross exposures have been used to add contagion channels to stress testing. The IMF’s introductory stress testing kit by Čihák (2007) includes a simple feature analyzing how a failure of a bank may affect other banks directly if it defaults on its borrowings. Espinosa-Vega and Solé (2010) and Tressel (2010) incorporate an additional channel: the failure of a bank may affect other banks indirectly because it stops the failed bank from lending to the other banks, thus eliminating a source of liquidity in the system. Such contagion analysis has been part of the joint Early Warning Exercise of the IMF and Financial Stability Board (with a focus on cross-border contagion using cross-border bank exposure data from the Bank of Interna-tional Settlements) and stress testing in Financial Sector Assessment Programs. Network effects can be incorporated into stress tests in an ad hoc manner (selecting randomly the bank[s] that fail, or “trigger” banks) or incorporated into the main (balance-sheet-based) stress testing exercise, where the institution(s) failing the solvency or liquidity tests under a particular scenario become the “trigger banks.”

Advanced network models can analyze further dimensions of networks. These models typically examine four measures of intercon-nectedness, depending on the structure of cross-exposure links, to identify key nodes (institutions, financial systems, or entire countries) in the network: (1) “in-degree,” which is the number of links (or vertices) that point to a particular node; (2) “closeness,” which is the inverse of the average distance from one node to others; (3) “betweenness,” which focuses on the shortest path between nodes; and (4) “prestige,” which assigns increasing scores to nodes that are connected to other high-scoring nodes. When applied to financial data, these models can, among other things, describe the importance of financial centers in transmitting shocks around the world or identify banks that may be small but can play a critical role in connecting financial centers.

Cluster analysis separates the network into subgroups (“clusters”) of nodes that have closer connections to each other than to those outside of the cluster. It can help identify subgroups of nodes with close connections and “gatekeeper” institutions or systems that bridge across different clusters (see Figure 2.2.1). This technique was used by the IMF to identify the 25 jurisdictions with the most systemically important financial sectors that are at the center of the global financial network (see Figure 2.2.2) (IMF 2010a). A similar technique was ap-plied by the Reserve Bank of India to identify core institutions, including banks and nonbanks (Figure 2.2.3).

These network models can help both model contagion directly and provide input to stress tests. For instance, the Reserve Bank of In-dia’s network model (Figure 2.2.3), based on the tiered network with a highly clustered central core, showed that a failure of a bank with large exposures to the insurance and mutual funds segments of the financial system could cause distress to 10 other institutions, including three insurance companies. Moreover, by identifying systemically important institutions at the core of the financial system, this approach can help set the right perimeter for stress testing.

Net PayableNet Receivable

In-CoreMid-Core Out-Core

PeripheryColor Code:

D015D034

D005 D004

D007D023D022

B011

D001

D024D019 D010

D003

D025B002

B003

B005

D033B014D029D032

B008

D026

D027

D013D009 B009 B001

A005

A003A001C007

A023D012

D017D006C001

D016A009A027A022A008D028

A016A007D021D014D012

A019A004 A025

D041

A013A020

Source: Reserve Bank of India 2011.Note: Round dots are banks (gray = net payable and blue = net receivable) and other shapes represent cooperative banks, asset management companies, insurance companies, and finance companies. The network has four distinct tiers, including the most connected banks (inner core circle of dots), mid-core, outer core, and periphery.

Figure 2.2.3. India: Network Structure of the Financial System (end-Sept. 2011)

Luxembourg

BelgiumJapanSpainFrance

SwitzerlandIndia Italy

GermanyIrelandTurkey

Russia Austria

SingaporeBrazil

Netherlands

ChinaHong KongMexico

CanadaKoreaUKUSA

Australia Sweden

Source: IMF 2010a.Note: Each round dot is a country, and blue lines represent cross exposures between countries. Triangles are the 25 most systemically important financial centers identified by applying cluster analysis to various measures of interconnectedness.

Figure 2.2.2. Position of the 25 Jurisdictions with Systemically Important Financial Sectors in the Global Network

GermanyLuxembourg

UnitedKingdom

BelgiumItaly

Tunisia

ChinaSpain

France

Poland

Algeria Malta

Switzerland

DominicanRepublic

Trinidadand Tobago

Venezuela

IrelandHungary

Bulgaria

Netherlands

United States

DenmarkSwedenFinland

Norway

Source: IMF 2012d.Note: Each entry is a country. Links between countries represent trade and financial linkages. Cluster technique is applied to identify subgroups (clusters) of countries with particularly strong links to each other (colored shapes including several countries).

Figure 2.2.1. Cluster Analysis and Global Trade and Financial Architecture



prehensive Capital Analysis and Review exercises in-clude a life insurance company (MetLife) and the former auto loan arm of General Motors (Ally), in ad-dition to investment and commercial banks, though these companies currently operate with bank holding company licenses.20 As the operator and overseer of the main payments system, some central banks use simulation tools to assess the impact of operational disruptions of the FMI itself or a major participant. Incidents are simulated to identify recovery times, critical participants, and contingency measures. Stress tests of central counterparties (CCPs) take into ac-count extreme but plausible market conditions, and are typically framed in terms of the number of partici-pant defaults that CCPs can withstand.

Principle 2 (Risk Transmission): Identify All Relevant Channels of Risk Propagation

In addition to network effects among financial intermediar-ies, there are other channels of shock propagation that relate financial intermediaries to each other and to other agents in the economy. Key examples of these propagation channels, illustrated in Figure 2.2, include:

• The feedback between liquidity and solvency risks (Wong and Hui 2009). This includes the (highly nonlinear)

important financial sectors that are required to un-dergo mandatory stability assessments under the FSAP (IMF 2010a).

• Practices are much more uneven for the inclusion of nonbank institutions in stress tests. FSAPs always test the banking sector, and occasionally the insurance sector; other sectors are rarely tested. In cases in which insurance is covered, effort is made to align the economic shocks to those used for the banking stress tests (Jobst, Sugimoto, and Broszeit 2014). But since each segment of the financial system may react differently to a certain macroeconomic scenario,19 the main scenario needs to be complemented with sector- specific scenarios. Moreover, insurers can also be vulnerable to structural and/or nonmacro risks (for example, catastrophic risks in the case of non- life insurers, or long- term changes in mortality, the case of life insurers), which are captured by single- factor shocks but tend to be rarely covered in FSAPs.

• Among country authorities, about 40 percent of the re-spondents indicate they only test the banking sector and another 45 percent also test the insurance sector. Tests of pension funds and FMIs are undertaken on a much more ad hoc basis, if at all. The US SCAP and Com-

19 For instance, banks and insurers often react to interest rate shocks dif-ferently, as banks typically have positive duration gaps, losing from ris-ing rates, but insurers have negative duration gaps, gaining from higher rates.

Shocks/Scenarios

Macroeconomic/Financial

Mark-to-market losseson assets available for sale

Trading income lossesNon-interest income losses

Credit losses

Asset side(market

liquidity risk)

Liability side(funding

liquidity risk)

Systemic Assets/Loss Distributions

Effects on banklending

Policy reactions

Effects onfunding and

capital markets

Feedback

Source: Authors.

Figure 2.2. Transmission Channels of Shocks between the Real Economy and the Financial Sector

20 The Dodd- Frank Act allows the US Federal Reserve Board to use its discretion to include banks and nonbanks in the stress tests.



on the duration of a crisis. Aït- Sahalia and others (2012) show the importance of both the right choice and timing of policy interventions to be effective during the time of a crisis.

The design of stress tests requires a careful examination of transmission channels and a sound understanding of the range of possible reactions of financial institutions and capi-tal market behavior to different shocks. While progress is being made on all of these fronts, there are still gaps in our understanding of the interaction between the real economy with the financial sector— the macro- financial framework— and of the role of the FMI and business practices in amplify-ing and transmitting negative shocks.

The operational implementation of this principle remains a major challenge, especially when it comes to the feedback effects between the real and financial sectors. As noted pre-viously, reliable stress tests require the identification and calibration of propagation channels (with historical infor-mation or expert judgment) and their incorporation in stress test design and implementation in the face of incomplete information, including dealing with the tail risk arising from “unknown unknowns.” Progress is being made rapidly on models addressing some of the transmission channels highlighted in Box 2.3. Although recent research has

relation between funding costs and risk perception by bank creditors and the relation between asset sales motivated by banks’ reactions to liquidity problems (fire sales) and further declines in asset values, which in turn aggravate solvency problems.

• The feedback from financial stress to the real economy (Krznar and Matheson 2017). Bank reactions to fi-nancial stress (for example, deleveraging and capital flight) can have adverse effects on the real activity, which in turn further degrades bank asset values ( second- round effects).

• The feedback between financial sector stress and sover-eign risk (Gray and Jobst 2010). Traditionally, the identification of risk from the relation between fi-nancial activities and the sovereign focused on the magnitude of contingent liabilities from implicit (or explicit) government guarantees to the banking sec-tor (and the financial system more generally). How-ever, the European sovereign debt crisis between 2010 and 2011 demonstrated that this relation is much more complex. This is discussed in more detail in Appendix 2.4.

• Policy feedback. Policy reactions (or lack thereof) can have a significant impact on risk transmission and

Box 2.3. Integrating Liquidity and Solvency Risks and Bank Reactions in Stress Tests

Banks have numerous ways to react to credit and funding shocks. High-quality capital and profits are usually the first line of defense, and retained earnings can help buffer banks’ capital levels. Banks have an inherent capacity to generate liquid assets by using high-quality eli-gible securities as collateral for market or central bank funding if interbank markets freeze. As seen post-Lehman, fire sales of securities are also an option, but at considerable cost in an environment of sharply declining asset prices. Deleveraging, especially targeted at assets with higher risk weights, is a way to raise capital adequacy ratios by reducing risk-weighted assets. In practice, banks have been using a combination of these, as well as other hybrid measures, ranging from debt-to-equity conversions to issuance of convertible bonds to opti-mizing risk-weighted assets, to react to shocks.

Incorporating banks’ reactions to shocks is a critical input into the design of informative stress tests, especially over longer time hori-zons. This, however, requires modeling solvency and liquidity shocks in a coherent manner because first, when banks react to financial stress, the source of the shock (solvency or liquidity) is not always clear; and second, the measures banks take in reaction to these shocks have both capital and liquidity aspects that are not easy to disentangle.

A relatively simple (but somewhat ad hoc) way to integrate solvency and liquidity shocks is to conduct two-round stress tests, with a bottom-up first round and a top-down second round. If, for example, most banks report the sale of particular asset classes in response to the shock in the bottom-up first-round test, the top-down second-round test could impose “haircuts” on those assets; if banks report that they would discontinue reverse repos, the analysis would incorporate a reduction in repo rollovers. The quantification of these haircuts or rollover rates could be based on historical information, cross-country experience, and/or expert judgment.

Several analytical approaches have attempted to integrate solvency and liquidity more systematically:• Schmieder and others (2012) simulate the increase in funding costs resulting from a change in solvency, indicated by a change in a

bank’s (implied) rating. • The Dutch Central Bank developed a stress testing model that tries to endogenize market and funding liquidity risk by including

feedback mechanisms that capture both behavioral and reputational effects. Several central banks and bank supervisors have been successfully using this framework.

• The Hong Kong Monetary Authority sought explicitly to capture the link between default risk and deposit outflows. Their framework allows simulating the impact of mark-to-market losses on banks’ solvency positions, leading to deposit outflows, asset fire sales by banks, and a consequent sharp increase in contingent liquidity risk.

• Barnhill and Schumacher (2011) developed a more general empirical model, incorporating the previous two approaches that attempt to be more comprehensive in terms of the source of the solvency shocks and compute the longer-term impact of funding shocks.

• Another attempt to integrate funding liquidity risks and solvency risk is the Risk Assessment Model for Systemic Institutions devel-oped by the Bank of England. The framework simulates banks’ liquidity positions conditional on their capitalization under stress, and other relevant dimensions such as a decrease in confidence among market participants under stress.



banking sector deleveraging and declining credit growth, are not captured. This area would require a new generation of macroeconomic models that in-clude a fully specified financial sector. Such research has been spurred by the crisis, in both policy institu-tions and academia, but it will take some time before sufficiently operational models are developed.22

• Network analysis or market- based systemic risk mea-sures are increasingly used for the analysis of spillover effects. The FSAP for Luxembourg (IMF 2011c) ex-amined network effects including bank- by- bank cross- border exposures, including exposures to par-ents and subsidiaries in the same financial group. Among country authorities, stress testing models such as the Bank of England’s Risk Assessment Model for Systemic Institutions model add conta-gion effects using network models. Market- based ap-proaches (see Box 2.1) reflect interconnectedness across institutions in a reduced- form manner in a process of estimating systemic losses accounting for interdependence among institutions, for instance, in the case of the FSAP for the United States (IMF 2010a).

Principle 3 (Scope): Material Risks and Buffers

Capturing all quantifiable risks is key to obtaining reliable stress test results. Until the global financial crisis, stress tests typically focused on credit risk from customer loans and market risk from marketable securities. The crisis revealed that this coverage was incomplete, and other sources of risk, such as sovereign, funding, systemic liquidity, and counter-party risks, should also be included in the stress tests to widen the coverage of potential sources of shock. For inter-nationally active financial institutions, incorporating cross- border exposures, credit and market risk of off- balance- sheet positions, and funding (including parent- subsidiary funding and liquidity transfers) is important for both home and host country authorities to assess the full risk profiles of the fi-nancial institutions.

Nevertheless, there are limits to the scope of risk factors to be included in stress tests. Some risks (for example, own sovereign) are so large and hard to hedge that financial insti-tutions and— especially— entire systems are likely to be very vulnerable to them, and any risk reduction or mitigation (for example, through additional capital) is bound to be so costly as to be infeasible. Similar issues arise for system- wide li-

enhanced our understanding of amplification mechanisms, such as those arising from funding shocks (for example, the adverse liquidity spiral as described in Brunnermeier and Pedersen [2009]), their incorporation in actual stress tests is difficult without a fully specified macro- financial model. Thus, some of the more salient transmission channels and effects are sometimes integrated in an ad hoc manner, by adding a second- round top- down stress test to bottom- up, bank- by- bank stress tests to assess potential bank reactions to the first- round shock. For example, a possible identifica-tion strategy could be informed by discussions with financial institutions about their responses to certain stresses (for ex-ample, portfolio allocation, deleveraging, and/or access to central bank funding). Such responses could help improve the design of stress test exercise, for example, designing the second- round shock by financial stability authorities. Against this background, it is no surprise that stress testing practice generally falls short in this area:

• Most country authorities that responded to the survey (IMF 2012c) and almost all FSAPs incorporate liquid-ity shocks in their stress tests, typically assuming the withdrawal of deposits and, in many cases, reductions in interbank exposures and access to secured funding (due to “haircuts” on liquid assets). In addition, in a few cases, countries also account for liquidity needs from off- balance positions and withdrawals of other types of wholesale funding. About half of the re-spondents consider domestic currency and foreign currency liquidity positions separately. However, most of these liquidity stress tests are implemented independent of solvency tests, excluding the possibil-ity of experiencing a more severe run on liabilities when banks are likely to make substantial losses (il-lustrated in the case of Bear Stearns and the experi-ence of some banks in the European periphery). Most large European banks compute their maxi-mum risk tolerance (ECB 2008)21 utilizing a sto-chastic approach, which aims at estimating their “ liquidity- at- risk” (maximum liquidity gap within a certain time horizon and for a given confidence level) or their “liquidity value- at- risk” (maximum cost of liquidity under certain assumptions). These approaches comprise single factor shocks and do not incorporate traditional credit risk stress or links to macroeconomic scenarios.

• Feedback effects from the financial sector to the macro-economy or the complex bank- sovereign linkages are rarely incorporated. In some FSAPs, fiscal- financial linkages have been explicitly modeled by incorporat-ing findings from fiscal debt sustainability analysis on the valuation of bank exposures and funding costs (IMF 2012a), or by relying on market- price- based ap-proaches (Appendix 2.4). But other types of feedback effects, such as the mutually reinforcing effect of

21 See also Matz and Neu 2006.

22 Two examples of the former include the Global Financial Stability Re-port (GFSR) (IMF 2011b), which presents a dynamic stochastic general equilibrium model with a banking sector and an econometric model incorporating the interactions between the financial sector and macro-economy developed by the staff at the Bank of Japan (Ishikawa and others 2012). The IMF’s Research Department has also expanded its global dynamic stochastic general equilibrium model— Global Inte-grated Monetary and Fiscal model— to explicitly include the financial sector.



and potential fees and commissions in a stress scenario can mask the impact of losses. Therefore, it is important to prop-erly understand the business conditions under stress. Sepa-rately, test results could differ depending on how the impact of stress scenarios on RWAs is incorporated; as with income, changes in RWAs can affect the stress test results, in some cases substantially.

Modeling the transmission channels between macroeco-nomic stress and nonimpaired components of income is a challenging task. It is particularly difficult for top- down ex-ercises due to (1) the lack of sufficiently granular information (for example, which assets and liabilities are based on fixed or floating interest rates, maturities, and banks’ hedging practices); (2) the complexity of the sources of bank income; and (3) the likely changes in banks’ behavior under stress as they protect their balance sheets (for example, many banks attempt to shield profitability through higher fees and com-missions when entering a downturn). Box 2.4 describes ap-proaches used in FSAP stress tests to tackle these challenges. In bottom- up tests, banks have many degrees of freedom, partly because there is no widely accepted single model for parameterization for nonimpaired profits. Careful examina-tion of individual banks’ models and comparison of assump-tions and results across institutions with similar business models should thus be an integral part of any stress testing exercise.

quidity shocks. This has led many to question the usefulness of stress testing these types of risk. Other risks have simply fallen out of fashion: for example, in an environment of low and stable inflation and interest rates due to more effective monetary policy regimes, refinancing and reinvestment risks arising from sudden asset- liability mismatches due to inter-est rate shocks, are often neglected. Regardless of the valid-ity of some of these arguments, the incorporation of all risks in a stress test remains an issue of paramount importance to obtain a complete picture and guide the search for risk- mitigating solutions. Not all potential risks need to be ad-dressed with additional capital, while comprehensive and candid stress tests could help gauge the consequences of in-action or delay (for example, in addressing sovereign debt sustainability).

Preimpairment income represents an important buffer to the impact of risk factors contributing to losses and should also be part of stress tests. Most stress tests in FSAPs cover risk horizons of two years or longer. Over such a time pe-riod, profitability can significantly influence the stress test results. For instance, in many of the prudential stress tests before the crisis projected preimpairment profits exceeded estimated losses; and in the first EU system- wide stress tests, pre impairment profits absorbed the impairment losses due to credit and market risks. However, overly optimistic ex-pectations about net interest margins, investment income,

Box 2.4. The Projection of Preimpairment Income under Adverse Scenarios

Stress test results depend not only on shock-related capital losses but also on preimpairment income (profit before losses from lending and investments), which contributes to capital. This factor is particularly important to consider in stress tests with longer risk horizons. In many cases, the size of preimpairment income can be substantial compared to the shock-related losses. But projecting preimpairment in-come is complex, and there is no widely agreed methodology to do so. Instead, stress testers employ a range of techniques.

The most straightforward way to project bank profit in line with macroeconomic conditions is to estimate the elasticity of preimpair-ment income (relative to capital) to economic growth. Hardy and Schmieder (2013) estimate this elasticity using BankFocus data for more than 16,000 banks. They find a decline of GDP growth by one percentage point reduces the ratio between preimpairment income and capital by between 1 and 1.5 percentage points. The average “preimpairment-income-to-capital ratio” in both advanced and developing economies is about 12 to 15 percent, and a moderate stress event (a 4 [6] percentage point drop in annual GDP growth in advanced [devel-oping] economies) would reduce it by about half. However, under severe stress conditions, the elasticity increases to a factor of 4 in ad-vanced economies, implying a strongly nonlinear relation between macroeconomic conditions and bank profits.

Depending on data availability, the main components of preimpairment income—net interest income, fee and commission income, and noninterest expenses (including salaries)—may be estimated separately. Net interest income could be projected in line with the inter-est rate assumption included in the macroeconomic scenario and assumptions for interest rate pass-through to bank borrowers. Hardy and Schmieder (2013) find that net interest income is more closely linked to macroeconomic conditions than other sources of income (such as fees and commissions), except for trading income under highly adverse conditions. There are also several studies (Ennis, Fessenden, and Walter 2016; IMF 2013b, 2017) that document a large influence of the level and slope of yield curve on banks’ net interest margin. Fees and commission income could be projected by examining their sources: if the majority is related to the sales and trading of securities, it is rea-sonable to project them in line with asset prices; if fees are mostly related to loan origination and credit cards, they should be linked to credit growth or employment.

The estimation of net interest income under stress could also incorporate interactions between solvency and liquidity stresses. When depositors and wholesale fund providers suspect that a bank may incur substantial losses, they would tend to reduce their exposure and/or raise the cost of funding due to higher counterparty risk. Higher funding costs, if not sufficiently passed onto borrowers, could reduce net interest margins. This is particularly true for wholesale funding costs, which are much more elastic than deposits (since at least part of the depositors’ funds are protected by deposit insurance) and harder to pass on to borrowers in weak economic conditions. Schmieder and others (2012) and Jobst (2014) have therefore constructed a method linking a bank’s funding cost to its solvency condition by using struc-tural models for credit spreads (similar to Moody’s KMV). While funding costs remain relatively stable for well-capitalized banks, they rise sharply once banks approach the minimum regulatory capital requirement. Depending on how much of the higher funding costs banks can pass on, their income (that is, net interest income) can shrink substantially.



brated based on historical data, as discussed under Principles 5 and 6).

• Stress tests are expected to complement regular prudential oversight. Since stress tests should primarily reflect the economic impact of certain shocks to the capital as-sessment, reported prestress capital adequacy ratios might need to be adjusted for potential underprovisioning, weaknesses in loan classification, collat-eral overvaluation, forbearance, and concentration risk from large exposures. However, these adjustments are judgmental and often involve large estimation er-rors, and should therefore be cross- checked with the expert judgment of the supervisory authority.

• For system- wide exercises involving banks with differ-ent business models, it is challenging to ensure method-ological consistency in projecting preimpairment income and capital intensity of exposures (that is, unexpected losses) without dismissing each institution’s idiosyn-cratic characteristics. In bottom- up exercises, the su-pervisory authority or the FSAP team typically attempt to impose some harmonization by enforcing uniform assumptions on preimpairment income (for example, interest pass- through). However, the im-pact of a shock on RWAs can vary across countries and institutions, partly reflecting differences in reg-ulatory treatment.25

• Some risks are often neglected, such as the impact of downgrades (that is, capital requirements for credit risk are based on default probabilities only and do not include capital buffers for downgrading risk) and interest rate risks in the banking book. Incorporating risks from cross- border exposures fully could be a challenge even if their importance is recognized in both home and host countries, due to (1) data issues as home- host col-laboration becomes critical; and (2) the need to con-struct global macroscenarios, which might be outside the scope of macroeconomic modeling teams of coun-try authorities. Some more recent FSAPs managed these difficulties by relying on bottom- up exercises and utilizing the IMF’s global macroeconomic frame-work to provide global macroscenarios.

Principle 4 (Interpretation): Make Use of the Investor’s Viewpoint in the Design of Stress Tests

Market perceptions of solvency and asset values matter for the design of the stress test. Prior to the global financial

The implementation of this principle requires a thorough understanding of the key transmission channels between risks and business conditions affecting bank portfolios in or-der to support a credible capital assessment under stress. This would enable the stress tester to capture all relevant risks and incorporate buffers appropriately in the design of the exercise. It also requires taking as comprehensive an approach to risks as possible, even if some of them may be “too big to mitigate.” Finally, this principle also provides an argument in favor of bottom- up over top- down tests to the extent that stress tests conducted by the banks themselves (with adequate supervi-sory scrutiny) may be more comprehensive and informative, since banks know better their portfolios and risks.

The experience of recent crises has spurred improvements in the way stress tests incorporate all material risks and buf-fers, but important gaps remain:

• The global financial crisis and the European sovereign debt crisis, as discussed in Section 3, illustrate that many of the relevant risk factors were not sufficiently addressed in stress testing exercises. Since then, the cov-erage of stress tests has expanded to include addi-tional risk factors and transmission channels, often with greater specificity. Before the crises, stress tests focused on shocks to credit risk and asset prices (mainly equity prices and real estate) as well as li-quidity and funding risks (IMF 2012c).23 After the crisis, liquidity and funding risks have gained greater prominence, and new risks emerged as im-portant elements of stress test scenarios, including sovereign risk, regulatory risks, and spillover/conta-gion risks due to interconnectedness.

• Sovereign risks have increasingly relevant in both FSAP and national authorities’ tests. System- wide stress tests in both Europe and the United States (EBA 2012, 2016, 2014; Board of Governors of the US Federal Re-serve System 2012) as well as FSAP stress tests in ad-vanced economies have covered sovereign risk since 2011. In most cases, these are modeled as expected losses from market risks by applying valuation hair-cuts on sovereign debt securities.24 However, the methodologies differ across exercises, especially re-garding the coverage of exposures (for example, stressing HtM exposures to own sovereigns, which is discussed more under Principle 4) and the size of the shock (which could be too small if shocks are cali-

23 Ong and Čihák (2010) discuss how the precrisis stress tests on Iceland generated deceptively benign results by ignoring liquidity risk (deposit withdrawal). They illustrate that liquidity stress tests using detailed pre-crisis disclosure information on funding positions could have indicated a vulnerability that materialized later.

24 Sovereign risk is ultimately a specific kind of credit risk. However, “sov-ereign risks” for advanced economies typically means sovereign market or spread risks, which are (unrealized) MtM valuation losses upon changes in market prices of sovereign bonds rather than outright de-fault risks (Hannoun 2011).

25 A stress scenario would not only affect capital (numerator) but also RWA (denominator). Survey results indicate that practices in calculating the latter vary (Le Leslé and Avramova 2012). Banks under the advanced internal ratings- based approach ( A- IRB) calculate their risk-weighted assets (RWAs) using borrower- or loan- specific probability of default/loss given default data following the Basel formulae. A deterioration in probability of default/loss given default leads to increases in RWAs. In contrast, banks under the stan-dardized approach apply specific risk weights depending on the types of ex-posures, and deterioration in loan quality does not necessarily affect RWAs.



economic capital (and provide a more conservative basis for the capital assessment under stress).

• Stressing market risk appetite. Market risk appetite can be explicitly stressed when designing stress test scenarios. For example, extreme adverse scenarios could incorporate an increase in the market price of risk similar to that experienced during a severe global crisis— say, of the magnitude of the 2008–09 crisis. This historical scenario would have an important impact on the bank’s risk- adjusted balance sheet, leading to higher credit spreads and funding costs, higher potential losses to bank creditors, and lower equity values. If these effects are not accounted for, the distribution of bank losses under an adverse sce-nario may be underestimated and lead to an overly optimistic assessment of bank capital during stress.27

• Imposing hurdle rates based on targeted funding costs. Hurdle rates based on regulatory ratios reflect what regulators consider an adequate solvency ratio, but the markets’ assessment of a bank’s solvency may be different. And markets may demand— and banks would have an incentive to target— capital ratios that enable them to attain a certain risk rating and keep their funding costs under a certain ceiling. Using hurdle rates that reflect market views (in addition to the regulatory minima) in stress tests recognizes this simple but stark reality. Otherwise, if stress material-izes, banks that “passed” the stress test may fail the market- based hurdle rate of capital adequacy. Box 2.5 presents two approaches developed by the IMF staff to calculate market- based hurdle rates. For supervi-sory authorities or central banks, which typically have better access to bank- specific and market infor-mation, the identification of the risk- return trade-off as perceived by the market should be simpler. Impos-ing hurdle rates that may be higher than regulatory minima is particularly important for macropruden-tial stress tests. Financial institutions must be suffi-ciently capitalized not only to ensure their own viability in the event of a system- wide shock, but also to prevent from becoming sources of risk propaga-tion. This may well require higher capital than is necessary when a financial institution is considered on a stand- alone basis (OFR 2012).

This principle also has implications for the public disclo-sure of stress test results. Publishing stress test results can

crisis, banks started to rely more on short- term wholesale funding and less on insured deposits. During the global fi-nancial crisis, uncertainty about counterparty risk and the valuation of these largely unsecured funding instruments triggered confidence shocks that caused major bank distress. With a larger share of financial institutions’ liabilities be-coming uninsured, markets can effectively force a bank clo-sure amid rising sovereign risk. This can be done by penalizing a bank with higher funding costs and, at the limit, depriving the bank from additional funding altogether. This market- imposed discipline affects bank performance through higher losses, need to deleverage, and possible further impair-ments triggered by second- round effects.

The operational implication of this principle is that mar-ket views need to complement stress tests based on regula-tory and accounting standards. There are several ways to do this including:

• Adopting market- consistent valuation of all bank as-sets and liabilities under the baseline and adverse sce-narios. In some cases, this would mean adjusting the prevailing accounting approach for HtM expo-sures, namely amortized cost principle net of any impairment provision based on incurred loss. Oth-erwise securities are assumed to remain unaffected by market prices, and, thus, valuation changes would not impact the bank’s capital. While this may be true from an income perspective (to the ex-tent that these securities would not be sold in an adverse scenario and obligors do not default), it may not be true from a valuation perspective used by investors to assess bank’s risk profile. Given that funding costs are an important element in all stress tests, adopting a full MtM approach would provide a useful benchmark.

• Using economic rather than statutory capital as the ba-sis for stress tests. Economic capital is a concept that attempts to capture the bank’s underlying economic value, and may deviate substantially from statutory capital, calculated on the basis of the jurisdiction’s current regulatory and accounting standards. For in-stance, in the event of a substantial decline in asset prices, banks’ securities holdings could carry unreal-ized losses that are not fully reflected in regulatory or accounting capital.26 Similarly, loan provisioning may become inadequate once substantial drops in collateral valuation are fully accounted for.

• Using PIT parameters to measure expected and unex-pected losses. PIT parameters, as opposed to regula-tory approaches to capital measurement (which are typically based on TTC parameters), may be a better way to reflect investors’ assessments of measures of

26 This idea was implemented by the EU Capital Exercise (EBA 2012), which requires banks to value all HtM or AfS sovereign debt exposures to European Economic Area member countries to be valued at fair mar-ket value.

27 The market price of risk is the compensation that a risk- averse investor would require for getting into a risk position. Many stress test method-ologies use actual default probabilities (for example, historical defaults in an industry sector or nonperforming loans). But these approaches do not reflect the true price of risk and also ignore its volatility over time. Several papers have developed methodologies to convert historical de-fault probabilities into risk- adjusted default probabilities (also called risk- neutral default probabilities), for instance Espinoza and Segoviano 2011, among others. FSAPs for Israel and Sweden (IMF 2011e, 2012e) have included the market price of risk in stress tests.



stress tests, including, but not necessarily limited to, capital injections.

Actual stress testing practices generally fall short of this principle:

• Market- based hurdle rates, such as a market- based capi-tal adequacy ratio, are generally not used by country au-thorities, and only rarely used in FSAPs (IMF 2010a; IMF 2011a). However, following the crisis, funding costs are increasingly being modeled and incorpo-rated explicitly in stress tests. While this is an impor-tant improvement, it is not enough. As shown in Box 2.5, the relationship between funding costs and bank

help remove asymmetric information during periods of uncertainty and restore market confidence. Even in the case of stress tests undertaken for surveillance purposes during noncrisis periods, the public communication of their results could create awareness of risks, promote more realistic risk pricing, and enhance market discipline— which, in turn, could reduce the probability of future crises. However, for the publication of results to yield these benefits, stress tests need to be candid assessments of risk, explicit about the coverage and limitations, and the announcement of their results needs to be accompanied by measures that will convincingly redress any vulnerabilities unveiled by the

Box 2.5. Market-Based Hurdle Rates

As a complement to regulatory hurdle rates, market-based hurdle rates should be used in stress tests. This market-based hurdle rate can be calculated directly from the trade-off between banks’ capital ratios and their funding costs. For

example, Schmieder and others (2012) derive a nonlinear relationship between solvency (measured as the capital adequacy ratios under the internal ratings-based approach consistent with banks’ implied default probabilities) and funding costs for a sample of German banks based on the following steps:

• Funding costs (that is, bank interest expenses measured as the excess spread above the government bond rate) were estimated for an average German bank for 12 quarters during 2007–2009.

• German banks’ weighted average MKMV’s expected default frequency were plotted against the funding costs for the same 12 quarters.

• The expected default frequencies are translated into capitalization ratios using the internal ratings-based formula to establish a link between funding costs and capital adequacy ratios as shown in Figure 2.5.1 (the mapping includes an additional capital cushion of 2.5 percentage points, in line with empirical evidence).

The results suggest that a regulatory capital ratio of about 15 percent would be a good benchmark for stable funding costs, which would be equivalent to a common Tier 1 capital ratio of about 11 percent. Below this level, the funding costs are very elastic to a decline in capitalization.

Alternatively, the contingent claims analysis, using option pricing theory, can be used to recover the banks’ probability of default im-plied in bank stock prices, the market value of bank assets (that is, the value of assets adjusted by the banks’ default risk), and the market value of risky debt. The latter can also be thought of as the risk premium required by the bondholders to compensate for the expected losses on their claims. This information can be used to construct the market-based capital adequacy ratio (that is, market capitalization to market value of implied assets), which can be mapped onto the corresponding spread required by banks’ creditors. Figure 2.5.2 presents these results for a sample of French banks. Again, the relationship is nonlinear and provides a methodology to determine a market-based hurdle rate.

0

5

10

15

20

25

30

1000 200 300 400 500

Funding Costs (Spread aboveTreasury bills, basis points)

Econ

omic

Cap

ital R

atio

(Bas

el II

)

Sources: Schmieder and others 2012; and IMF staff calculations.Note: EDF = expected default frequency according to Moody’s KMV.

Figure 2.5.1. Basel II (EDF-Implied) Capital Ratio vs. Funding Costs for a Sample of German Banks, 2007–2009

0

8

1000 200 300 400 500

Five-year FV CDS spread,basis points

CCA

CAR

(per

cent

)

1

2

3

5

7

4

6

Sources: Bloomberg L.P.; Moody’s KMV; and authors.Note: CAR = capital adequacy ratio; CCA = contingent claims analysis; FV = fair value. CAR and spreads are (daily) median value for Crédit Agricole, Société Générale, BNP Paribas, and Natixis.

Figure 2.5.2. CCA CAR vs. Five-Year FV CDS Spread, for a Sample of French Banks, 2007–2009



happened. BCBS (2004) has provided some quantitative guidance for determining tail risks for single- factor shock stress tests.30 However, there is no comparable guidance on macroeconomic scenarios.

This concern is especially pressing when stress tests take place in crisis or near- crisis situations. In these cases, the fi-nancial institution or system is already experiencing signifi-cant distress. Thus, some supervisory authorities may be reluctant to apply excessively negative tail scenarios to an already stressed baseline projection. Moreover, publishing the outcome of stress tests that incorporate extreme scenar-ios in these circumstances could trigger self- fulfilling crises. Political economy and legal constraints could also come into play when choosing scenarios, especially when the results could serve as basis for determining the restructuring or resolution of a failing banks (or the determination of public sector support). On the other hand, as indicated by the con-trasting experience of SCAP and EBA tests (Appendix 2.3), compromising on the severity of scenarios could undermine the credibility of the exercise and prolong the crisis. These trade- offs are not easy to tackle. In principle, an effec-tive stress test in a near- crisis situation should not compro-mise on the severity of scenarios, but instead mitigate the possible adverse market impact of the results by having cred-ible support measures in place. If this is not feasible, it might be preferable not to conduct stress tests at all, but instead release critical exposure data and provide qualitative analysis to enhance transparency.

Another shortcoming of traditional approaches, regardless of the size of assumed shocks, is that they ignore the interde-pendence among shocks and among affected institutions or systems. For stress tests to provide a reliable assessment of the resilience of an institution or an entire financial system, they must consider not just the potential size of individual risk factors but also their dependence under all plausible scenar-ios. In addition, risk factors may be weakly correlated under normal economic circumstances but highly correlated in times of distress.31 Similarly, the joint default risk within a system varies over time and depends on the individual firms’ likelihood to cause and/or propagate shocks arising from in-dividual risk factors. Given that large shocks are transmitted across entities differently than small shocks, measuring non-linear dependencies in stress tests can deliver important in-sights about the joint tail risks.

But while it is relatively straightforward to generate stress test results based on the effect of a single risk on a static mea-sure of capital, combining multiple risks and the extent to

solvency is nonlinear, which suggests that there are some market- related solvency thresholds that can and should be used as hurdle rates.

• The assessment of sovereign risk is another area where market considerations are generally not incorporated in stress tests. Regulatory and accounting standards could substantially understate the valuation risks of sovereign securities when a significant share of these securities is held in HtM accounts. Most FSAPs now include unrealized losses from securities in HtM ac-counts by applying MtM valuations in response to yield changes, recognizing that it is the economic rather than the accounting valuation that matters for sustainability. But some national authorities or fi-nancial institutions have not embraced this ap-proach. However, the Bank of Japan regularly tests interest rate shocks on all securities, including those in HtM accounts, as part of its regular macropru-dential surveillance.

• Adjusting the initial capital for already realized shocks is often an integral part of supervisory and crisis man-agement exercises. For instance, the EU capital exer-cise (EBA 2012) required banks to mark to market all their sovereign bond holdings incorporating the large yield changes materialized in 2011. The stress test exercises conducted by Blackrock for Ireland and Greece for estimating recapitalization need included reassessment of existing loans by going through re-underwriting process. However, surveillance stress tests often do not include these adjustments.

Principle 5 (Calibration): Focus on Tail Risks

The rule of thumb for stress tests has traditionally been to apply “extreme but plausible” shocks, but there is no system-atic way to determine these. Typically, the size of shocks de-fined by a unit of measure or a qualitative attribute of severity, such as “ worst- in- a- decade” events, “1 percent prob-ability” tail events,28 or a “ x- times standard deviation” shock, which is calibrated to a historical scenario of one or more (macro) variables.29 One obvious problem with this ap-proach is that historical experience varies across countries and changes over time. A worst- in- a- decade scenario today would look very different than the same scenario in 2007. Another constraint, which became particularly relevant for stress tests conducted before the crisis, is the lack of suffi-cient historical price information on new financial products (for example, asset- backed securities) to calibrate adverse shocks. And obviously, an approach based on historical data would not work when considering an event that has never

28 If a 50 percent decline in equity prices happened only once in the past 100 years, it has statistical significance of 1 percent.

29 Assuming a normal distribution, a two- standard-deviation negative shock would have a 2.275 percent probability of occurrence, which can be thought of as approximately equivalent to a once- in- 50-year event.

30 BCBS (2004) suggests for G10 currencies either a ± 200- basis- point parallel rate shock or the first and 99th percentiles of observed interest rate changes using a one- year risk horizon period based on a minimum of five years of observations for calibration.

31 In these instances, there is a considerable shift of the average away from the median (“excess skewness”) and a narrower peak (“excess kurtosis”) of the probability distribution. If distributions become highly skewed, large tails may even cause the mean to become undefined, which is an important complication when using stress tests.



More recent FSAPs are making a systematic effort to in-clude severe shocks that are comparable across countries.32 One rule of thumb applied in many FSAPs is to assume a two- standard- deviation shock on GDP growth rate for two years based on a long (20–30 year) history, unless a different magnitude of shock (in either direction) is justified. Among country authorities, most survey respondents adopt tail events with small probability (ranging from 1–5 percent), while some use qualitative criteria, such as in line with, or worse than, historical worst (IMF 2012c). Since the global economy experienced extremely sharp economic deteriora-tion in 2008–09, shocks calibrated using a sample that in-cludes the crisis time observations should generate fairly severe shocks. Another frequently used approach is to test for a shock of similar magnitude as the global financial crisis. At the same time, all historical data- based approaches always carry the risk of complacency. This needs to be managed by more qualitative approaches (Principle 6).

Incorporating risk interdependence in stress tests contin-ues to remain a difficult technical challenge. Based on the survey results, several central banks and supervisory authori-ties attempt to capture risk interdependence heuristically by applying common macro and financial shocks and assessing interbank contagion through network models (IMF 2012c). The IMF staff has also been increasingly using network models on a stand- alone basis or as a part of a stress testing model. Beyond these heuristic approaches, some FSAPs have

which firms’ default risks may be correlated under different scenarios is very complex (Box 2.6):

• In the context of balance- sheet- based models, network models can help capture spillover effects and the proba-bility of many institutions defaulting (Box 2.2). How-ever, most conventional balance- sheet- based stress test models do not formally account for default de-pendencies across institutions. In this case, complex-ity arises from the amount of information necessary to identify the right network.

• In market- price- based models, it is possible to model joint default probabilities. In this case, the key challenge is modeling the dependence structure. One approach used frequently is estimating conditional VaR, which measures the VaR of the financial system conditional on the distress of one or more financial institutions in the system. The value of an institution and distress correlations among institutions are estimated using equity or other market price data. Another related ap-proach is the marginal expected shortfall (Acharya and others 2010) that specifies historically expected losses of a financial institution conditional on the fi-nancial system having breached some systemic risk threshold. Adjusting marginal expected shortfall by the degree of firm- specific leverage and capitalization yields the systemic expected shortfall, which yields the average, linear, bivariate dependence between banks when the entire banking sector is undercapitalized. More complex approaches require a significant depar-ture from conventional statistical methods (Box 2.6). Stress testing practices in this area are evolving rapidly in line with the development of new analytical tools.

Box 2.6. Risk Interdependence in System-Wide Stress Testing

For stress testing to inform a realistic assessment of capital adequacy, the objective is to consider both the variability of risk factors and their mutual dependence under all plausible scenarios (in addition to their likelihood and severity). While it is generally straightforward to generate stress test results based on the effect of a single risk on a static measure of capital, combining multiple risk factors under different scenarios—and on a system-wide basis—complicates a reliable capital assessment.

Estimating the dependence of risk factors to account for the low probability of joint (negative) extreme outcomes (with no or little his-torical precedent) is not straightforward and requires a significant departure from conventional statistics. The traditional correlation coef-ficient detects only linear dependence between two variables (or risk factors) whose marginal distributions are assumed to be normal. This statistical inference assumes an empirical relation (or the lack thereof) based on relatively more central (and more frequent) observations, and implies that the bivariate distribution of these variables is elliptical, which is hardly encountered in reality. Alternative measures of dependence between risk factors can capture the nonlinear dynamics of changes in variables far removed from the median. For example, an expedient nonparametric method of investigating the bivariate empirical relation between two random vectors is to ascertain the inci-dence of shared cases of cross-classified extremes via a refined quantile-based chi-square statistic of independence. Similarly, copula func-tions and other nonparametric methods provide the possibility of combining two or more distributions of variables based on a more flexible specification of their dependence structure at different levels of statistical significance. These approaches generate measures of so-called “joint asymptotic tail dependence,” which define the expectation of common extreme outcomes.

The measure of nonlinear dependence of risk factors can be combined with their individual severity distribution (or the risk profile of each bank) to derive a tail-sensitive estimate of joint default risk and the system-wide capitalization of multiple institutions under stress. For instance, closed-form methods under extreme value theory, such as the generalized Pareto distribution, are frequently used to help define the limiting behavior of different risks affecting the operating performance of a bank in extreme situations.1

(continued)

32 The effort is partly in response to the Independent Evaluation Office of the IMF assessment (IEO 2006), which recommended that FSAPs should aim for more severe stress scenarios applied in a similar manner across major countries.

1 Two prominent examples are the generalized extreme value distribution and the generalized Pareto distribution. While generalized extreme value helps model the asymptotic tail behavior of the order statistics of normalized maximums (or minimums) drawn from a sample of depen-dent random variables, the generalized Pareto distribution represents an exceedance function that measures the residual risk of extremes beyond a given threshold (that is, a designated maximum [or minimum]) as the conditional distribution of mean excess.



2 In contrast to the traditional (pairwise) correlation-based approach, this method links the univariate marginal distributions of expected losses in a way that formally captures both linear and nonlinear dependence in joint asymptotic tail behavior over time.

One example of this approach in IMF stress testing is the two-step estimation of joint expected losses within the Systemic Contingent Claims Analysis (Jobst and Gray 2013) framework. This alternative approach is fundamentally driven by modeling extreme dependence to quantify possible linkages between expected losses in support of a more comprehensive assessment of common vulnerabilities than would be allowed by balance-sheet-based approaches. After estimating the nonparametric dependence function of individual expected losses, it is combined with their marginal distributions, which are assumed to conform to a generalized extreme value (GEV) distribution, which identifies the asymptotic tail behavior of normalized extremes. The dependence function is estimated iteratively on a unit simplex that optimizes the coincidence of multiple series of cross-classified random variables. Finally, the conditional value at risk of joint expected losses is determined as the point estimate of the probability weighted residual density beyond a prespecified statistical confidence level.

The estimates of joint default risk under extreme scenarios can be used to cross-validate results from traditional, accounting-based stress testing approaches. In several Financial Sector Assessment Programs, most notably, in the cases of Germany, Sweden, the United Kingdom, and the United States (IMF 2010a, 2011d, 2011e, 2011f), institution-level stress tests of the banking sector were combined with systemic contingent claims analysis to derive capital assessments from a system-wide perspective—after controlling for the market- implied dependence structure of risk factors and their sensitivity to extreme events.

In general, incorporating time-varying extreme dependence of risk factors—ideally together with a market-derived measure of capital adequacy—offers a more realistic capital assessment in stress tests. This approach identifies the amount of capital shortfall to current capi-tal levels based on market expectations on solvency that far exceed the minimum regulatory requirements. For instance, the assumption of higher uncertainty about the realization of expected losses (that is, greater historical volatility) reflects the notion that, especially during distress periods, firms would need to have higher capital buffers in place to absorb the realization of losses above existing provisioning levels so that current capital levels remain unaffected over a specific risk horizon.

Box 2.6. continued

0

Time

Probability

Aggr

egat

e Lo

sses

Historical Volatility

(e.g., 1 standard deviation)

Shocked Volatility

(e.g., 2 standard deviations)

Expected drift of aggregate losses

(also informs historical loss scenarios)

ExtremeLoss

New ExpectedLoss (after shock)

Aggregate lossdistribution at time T

(after shock)

Additional regulatory capitalMean

Additional economic capitalVaR95 percent

VaR99 percent

T

0

Time

Probability

Aggr

egat

e Lo

sses


(e.g., 1 standard

deviation)


(also for historical loss scenarios)

ExtremeLoss

“tail risk”

Expected Loss (EL)= Minimum Regulatory Capital

Unexpected Loss (UL)= Economic Capital

Aggregate loss distributionat time T

Balance Sheet Approach(accounting values)

“Distributional Approaches”(estimated or simulated tail risk)

Value-at-Risk (VaR) orExpected Shortfall (ES)

Mean

VaR95 percentES95 percent

average densitybeyond VaR95 percent

VaR99%

T

Figure 2.6.1. Key Conceptual Differences in Loss Measurements—Implied Capital Requirement under “Distributional Approaches”

Sources: Jobst and Gray 2013; and authors.Note: VaR = value at risk.



Public disclosure may also create difficult trade-offs in some cases. Public disclosure of stress test methodologies, underlying exposures, assumptions, and results can help (1) raise public awareness of risks; (2) promote more realistic risk pricing and strengthen market discipline, thereby re-ducing the probability of future sudden reversals of inves-tors’ sentiment; and (3) inform more effective financial stability policies. Even when the results are weak, public communication can have a positive impact if accompanied by credible contingency plans and support measures for fi-nancial institutions that fail the tests, reflecting the authori-ties’ recognition of the problems and commitment to financial stability. At the same time, greater disclosure car-ries risks. It may (1) entice financial institutions to make portfolio choices to “game” the tests; (2) increase moral hazard and encourage complacency if investors rely exces-sively on published stress test results— which are always subject to a margin of error (Principle 7)—at the expense of other bank soundness indicators; and/or (3) undermine confidence if the necessary support measures are not in place (for political economy or other reasons). Also, as stress tests become more common, disclosure of different results can cause confusion— a problem encountered in FSAPs for EU countries when they took place concomitantly with EU system- wide stress tests.

Balancing the benefits and costs of disclosure depends partly on the circumstances and the nature of the tests. In a crisis situation, when market confidence is at a premium, a strong case can be made in favor of greater public disclosure of system- wide stress test results. In normal times, a more conservative approach might be warranted. Even then, how-ever, regular publication would familiarize market partici-pants to stress tests, making them a more effective tool at times of crisis. This is particularly true for system- wide stress tests (rather than microprudential stress tests as part of capital adequacy reviews), which are more likely to apply consistent assumptions across financial institutions, making the results comparable and guiding the attention on sys-temic risk factors that are relevant for maintaining or restor-ing market confidence.

Realizing the benefits of greater disclosure depends on a number of crucial preconditions. The stress tests should be credible. For this, they need to cover the relevant risks and transmission channels, assume serious shocks, set appropriate hurdle rates, and produce a candid assessment. And crucially, they should be accompanied by a convincing framework for follow- up action, including government support, if needed. If these preconditions are not met, disclosure would not be in-formative and might do more harm than good.

Lastly, the disclosure of stress tests should be set in the perspective of the broader communication strategy of finan-cial stability policies. Just as stress tests should be one of sev-eral assessment tools, so should their disclosure be part of a coherent overall communication strategy. Publishing stress test results is likely to be much more effective if done in the context of regular outreach aimed at informing markets and

tried to include some of these nonlinear dependences for-mally in the stress testing framework. In the FSAPs in Ger-many, Sweden, the United Kingdom, and the United States (IMF 2010a, 2011d, 2011e, 2011f), institution- level stress tests were combined with an attempt to derive capital assess-ments from a system- wide perspective, which considered losses from a portfolio of key financial institutions after con-trolling for the dependence structure of risk factors and the stochastic nature of input parameters. Technical limitations, however, mean that it will take some time before these ap-proaches are “mainstreamed” in FSAPs.

Principle 6 (Communication): When Communicating Stress Test Results, Speak Smarter, Not Just Louder

The experience of the global crisis underscored the importance of effective public communication of stress test results. Central banks and supervisory authorities in several countries pub-lished stress test results in their financial stability reports even before the crisis, at varying degrees of detail. FSSAs— most of which are published with the consent of national authorities— always include stress test assumptions and aggregate results (often supplemented by a more detailed description of data coverage and methodologies in Technical Notes). In the wake of the financial crisis, national authorities enhanced the disclo-sure of stress test results, especially in the United States and Europe, where policymakers saw it as a way to shore up market confidence. In the United States, the publication of the stress test results was established in national law.33 This spurred greater public interest in— and scrutiny of— stress tests, which, in turn, increased pressure for greater disclosure.

Public disclosure of stress tests can yield significant ben-efits but is no panacea. The US SCAP successfully restored market confidence in the banking sector, allowing investors to differentiate between banks based on their resilience, which arguably facilitated the raising of additional capital from private sources. In contrast, the 2011 EU system- wide stress test did not fully achieve this objective. This was not due to differences in the scope of public disclosure between the two exercises— in fact, the latter was praised for its trans-parency. Rather, it reflected differences in the design of the stress tests and the context within which their results were published. While the setup of the SCAP was credible (com-bined with a fiscal backstop), clearly communicated, and in place ahead of time, the European stress test was considered mild and not fully capturing the risk profile of weaker na-tional banking systems. And perhaps more importantly, the follow- up actions and policy backstops for failing banks were considered ambiguous (Appendix 2.3).34

33 The Dodd- Frank Act requires the Board of Governors of the US Federal Reserve System to disclose summary results of stress tests for large banks.

34 See Ong and Pazarbasioglu 2014.



in real estate deserve a stress test on the impact of a large de-cline in real estate prices regardless of the probability of such a shock. Alternatively, the US Federal Reserve Board typi-cally uses two scenarios for stress tests. One is unique to each institution and chosen by it; the other is common. In this way, the institutions are assessed under scenarios that they themselves consider particularly damaging.36 Reverse stress tests by individual institutions (Appendix 2.1) and surveys of such exercises across institutions could help extend the fron-tier of tail risks.

Another approach is the application of distribution the-ory to the scenarios themselves, as opposed to the current practice of choosing just one adverse scenario. This reflects the recognition that the future is stochastic and can be rep-resented by a number of event combinations, each of which with a probability of realization. A scenario distribution ap-proach was used by the IMF staff in the first FSAP for South Africa. Based on the statistical properties of the historical distribution of price changes, and using Monte Carlo simu-lations, each scenario was represented by a combination of changes in prices, including credit spreads, which were used to revalue bank assets. The final outcome represented a dis-tribution of bank capital ratios for each bank, in which each point of the distribution was associated with a particular scenario.37

Ultimately, the principle of being aware of the “black swan” is more about the context of stress test scenarios than about the mechanics of their design and implementation. It serves as a reminder that stress tests should not be under-taken in isolation and their results should not be taken too literally. No matter how much a stress tester tries, stress tests always have margins of error. Their results will almost always turn out to be optimistic or pessimistic after the fact. In ad-dition, there will always be model risk, imperfect data ac-cess, or underestimation of the severity of the shock. One should therefore set stress test results in a broader context.

Stress testing is just one of the many tools to assess key risks and vulnerabilities in financial institutions or entire systems. They should be treated as complements to other tools that can provide information about potential threats to financial stability, such as qualitative and quantitative bank risk analysis, early warning indicators, models of debt sus-tainability, and informed dialogue with supervisors and market participants, among others. Final conclusions about the resilience of the institution or system should draw on all these sources and not just on the results of stress tests.

the public about financial stability issues, and complemented by disclosure of a broad set of indicators.

The survey of actual practices confirms the trend toward greater disclosure (IMF 2012c):

• Most surveyed countries communicate the results of macroprudential stress tests (almost 85 percent in the case of solvency tests and 50 percent in the case of li-quidity tests). Usually, this takes place in annual (or semiannual) financial stability reports. In most cases, results are communicated using system aggre-gates, often with some distribution metrics, without disclosing the identity of individual institutions— with the notable exceptions of the EBA and US Fed-eral Reserve tests. FSAP stress test results are reported in FSSAs, which typically do not disclose the identity of individual institutions. Most FSSAs are published. Technical Notes on stress tests, which report much more detailed results, are increasingly published, especially in advanced economies.

• Raising public awareness of financial stability issues, achieving greater transparency, and providing informa-tion to market participants are mentioned as the main objectives of public communication. Although public communication is seen positively by all respondents who publish, many expressed concerns about the risk of exaggerated expectations placed on stress tests, in-consistent interpretation of stress tests by mass me-dia, and excessive focus by banks on published stress test results that could undermine their effectiveness as a supervisory tool.

Principle 7 (Limitations): Beware of the “Black Swan”35

There is always a risk that the “unthinkable” will materialize, regardless of how extensive the coverage of risk factors, how refined the analytical models, how severe the shocks incor-porated in the stress tests, and how careful the communica-tions strategy. Stress tests provide a measure of the resilience of a financial institution or a system to given shocks but can-not predict the future. However, future shocks might arise from new sources and unexpected events, which have historically shown little volatility or have not materialized for such a long time that they have been forgotten (for ex-ample, advanced country sovereign defaults). What practical ways are there to incorporate these factors into stress test design?

One approach is to design hypothetical scenarios based on expert judgment and new information, where available, rather than simply be guided by history. The Bank of En-gland (Haldane, Hall, and Pezzini 2007) proposed using cur-rent vulnerabilities as a guide for the choice of hypothetical shocks. This means, for example, that systems concentrated

35 The term “black swan” was first used by Taleb (2004) to indicate highly improbable events that have a major impact.

36 There are also policy trade- offs in designing shocks. On one hand, it is important to prevent a Type II error of allowing weak banks to pass the test. Dexia is a good example of a bank that passed the July 2011 EBA stress test but failed shortly afterwards. On the other hand, it is not use-ful to design shocks that are too severe, causing an excessive failure rate of banks that would be deemed sound under most adverse scenarios.

37 The methodology for this approach is discussed in Barnhill and others 2002, which is similar to Borio, Drehmann, and Tsatsaronis 2014, which suggests that stress tests themselves need to be stress tested by assessing the sensitivity of results to changes in assumptions.



hurdle rates. These decisions are key for the effectiveness of stress tests and the reliability of their results. When pressures from the industry (especially in the case of bottom- up tests) or political economy or other considerations unduly influ-ence these decisions, stress tests can do more harm than good. Deriving benign conclusions from stress tests that as-sume modest shocks, include optimistic projections of fu-ture income, and use trivial hurdle rates can lead to complacency. Disseminating publicly such results could un-dermine the credibility of the exercise.

However, the success of stress tests cannot be reduced to the choice of a few parameters but should be seen in the broader context outlined by the proposed principles. Stress tests are complex exercises with many “moving parts.” Their effectiveness does not depend on just a few parameters or on the degree of public disclosure of their results but also on the context within which they are conducted. This context in-cludes (1) a clear understanding of the stress tests’ scope and objectives; (2) knowledge of the key individual financial in-stitutions in the system, their business models, and main channels of risk transmission; (3) appropriate decisions on the tests’ perimeter and coverage; (4) consideration of complementary assessment tools; (5) a communications strategy tailored to the circumstances and purpose of the tests; and (6) a credible commitment to take the measures that may be required to address vulnerabilities uncovered by the tests.

Adherence to these principles is uneven in practice. Table 2.4 summarizes key operational implications that flow from each principle. For Principles 1–3, the recommendations focus on the preparatory work that needs to be conducted by the stress tester to conduct tests with adequate coverage of in-stitutions, risks, and channels of risk transmission. For Prin-ciples 4–7, the recommendations focus on stress test design and communication. A survey of central banks and supervi-sory authorities in 23 countries and stress tests in FSAPs showed that despite major improvements since the crisis, prac-tices still fall short of these principles (IMF 2012c; Jobst, Ong, and Schmieder 2013). Shortcomings are particularly notable in three areas, which reflects both gaps in the analytical tool-kit and weaknesses in implementation: (1) identifying the channels of risk propagation, (2) using the investors’ view-point, and (3) focusing on tail risks. The proper interpretation of stress test requires a sufficient understanding of the implica-tions of these gaps, which the current generation of stress test-ing exercises aims to close with priority.

Stress test design, models, and implementation should be “ back- tested” to the extent possible and regularly reassessed. Back- testing can take the form of a comparison of stress test outcomes under baseline scenarios with actual outcomes. For adverse scenarios, one could have stress test results re-viewed by a panel of external experts to assess their rationale and consistency of results across banks. Checking the ro-bustness of the results for variations in key parameters (in other words, stress testing the stress test), assessing the im-pact of new tools and new approaches and, last but not least, remaining vigilant for the emergence of new risks are crucial to ensuring more reliable tests.

5. CONCLUSIONS AND OPERATIONAL IMPLICATIONSThis chapter contributes to the debate on stress test design and implementation by proposing a set of operational best practice principles. The wide variety of stress testing ap-proaches and underlying assumptions have raised important questions about the interpretation of the test results and their comparability. Setting best practice principles and en-suring adherence to these could improve the integrity of stress testing exercises, promote greater transparency and comparability of stress tests across countries and over time, and ultimately contribute to more meaningful and effective financial stability assessments.

A key goal of this chapter was to set realistic expectations about what stress tests can and cannot accomplish. Stress tests are forward- looking tools to assess financial institu-tions’ solvency and liquidity and the resilience of the entire financial system under possible adverse scenarios, but they do not predict the likelihood of these scenarios materializ-ing. As such, regardless of refinements and improvements, they will always remain hypothetical statements. One should therefore always be cautious about using stress test results in isolation: a well- rounded risk assessment should use stress tests in conjunction with other tools to broaden the under-standing of vulnerabilities.

The discussion highlighted several important decisions in the design and implementation of stress tests. These involve the choice of (1) risk scenarios— in terms of both the coverage of all relevant risk factors and their severity, (2) types of tests so that they cover all relevant transmission channels and in-clude realistic assumptions about buffers, and (3) appropriate



TABLE 2.4

Practical Implications of “Best Practice” Principles for Stress Testers

Principle 1. Define the appropriate institutional perimeter.

• For system-wide stress tests, if universal coverage is not an option, identify systemically important institutions to be covered in the tests, including relevant nonbanks and financial market infrastructures. Criteria for assessing systemic importance should include an assessment of interconnectedness.

• Gain a basic understanding of the structure of financial conglomerates, if present, and cover in the tests any of their banking or nonbanking activities that may have a significant impact on financial stability during times of stress.

Principle 2. Identify all relevant channels of risk propagation.

• Identify and understand the main channels of risk propagation, including but not limited to: (1) the relation between solvency and liquidity conditions, (2) risk transfers between banks and the sovereign, (3) potential bank reactions to stress, and (4) feedback effects between financial sector stress and the macroeconomy (for example, through deleveraging and lower GDP growth).

• Incorporate as many of these channels in the design of the stress tests as possible. If modeling them explicitly is not feasible, use heuristic means, such as a second-round top-down test that maps the impact of first-round tests on individual institutions onto key macro-financial variables (asset prices, credit growth, aggregate liquidity) or expert judgment, to capture all potential channels of risk transmission.

Principle 3. Include all material risks and buffers.

• Understand the tested institution’s—or, for system-wide tests, the systemically important institutions’—business model, including markets where they operate and their sectoral or cross-border exposures.

• Be as comprehensive as possible in including potential sources of risk in the tests. Do not omit risks because they may be “too big to mitigate.”

• Assess whether any banks may be exposed to risks from cross-border exposures, non-banking activities (such as employee pension funds and investment funds) or other institutions that may be legally separate but with significant economic links (such as parent companies, subsidiaries, and off-balance-sheet vehicles), and include as many of these as possible in the tests.

• If stress tests include non-banking institutions, consider relevant noneconomic risks that may have a major impact on them and the rest of the system (such as natural disasters for insurance companies).

• In stress tests with longer time horizons, model explicitly likely buffers, such as preimpairment income. Preimpairment income should be scenario-dependent and conservatively projected.

Principle 4. Make use of the investors’ viewpoint in the design of stress tests.

• Adopt market-consistent valuation of bank assets and liabilities and point-in-time parameters to measure expected and unexpected losses. • Adjust institutions’ initial capital for supervisory practices or other factors that result in systematic over-reporting of capital, such as

underprovisioning, collateral overvaluation, and forbearance. Use economic capital estimates as a complement to statutory capital in the stress tests.

• Stress market risk appetite by including explicitly shocks involving increases in the market price of risk. • Use hurdle rates reflecting market views (for example, target funding costs) in addition to regulatory minima.

Principle 5. Focus on tail risks.

• Use a variety of approaches to determine “extreme but plausible” shocks to be used in stress tests, including calibration on historical data, cross-country experience, and “worst-ever” events.

• Whenever possible, adopt methodologies that capture tail events and risk dependencies. Ideally, use both balance-sheet-based models and market-price-based models, as the latter can incorporate more easily measures of risk dependencies using market prices. If using balance-sheet-based models alone, supplement stress tests of individual financial institutions with tools that measure spillovers and joint default probabilities, such as network models.

Principle 6. When communicating stress test results, speak smarter, not just louder.

• Tailor the communication of stress test assumptions, methodologies, and results to circumstances and the goals of the tests. Disclosure can yield substantial benefits for macroprudential stress tests, especially in crisis situations, when market confidence is at a premium.

• Before disclosing stress test results, ensure that the tests cover relevant risks and transmission channels, assume serious shocks, set appropriate hurdle rates, produce a candid assessment, and are accompanied by a credible framework of follow-up action, including government support, if needed.

• To the extent possible, set the communication of stress test results in the context of a broader communication strategy for financial stability objectives and policies, and complement them by other financial soundness indicators.

Principle 7. Beware of the “black swan.”

• Supplement statistical approaches to the identification of potential tail events with expert judgment and new information. Consider analyzing the sensitivity of results to variations in the assumed shock, instead of choosing just one adverse scenario.

• Supplement stress tests with other assessment techniques, such as qualitative and quantitative bank risk analysis, early warning indicators, models of debt sustainability, and informed dialogue with supervisors and market participants, among others. Final conclusions on the resilience of an institution or a system should be based on all these sources, not just the results of stress tests.

• Evaluate and regularly reassess the institutional coverage, risks, and channels of risk transmission included in the tests as well as the models and other approaches. Constantly assess the suitability of new tools. Remain vigilant for the emergence of new risks.

Source: Authors.


Appendix 2.1.Existing Supervisory Guidelines for

Stress Testing by Banks

The Basel III framework (BCBS 2017a) requires banks to maintain rigorous stress testing programs. For instance, EU banks that apply the internal ratings- based approach conduct regular stress tests to determine the regulatory capital for market risk under Pillar 1 of the Capital Requirements Directive, which applies the Basel framework to the European context. Internal ratings- based banks are further asked to run credit risk stress tests in order to examine the robustness of their internal approach. How-ever, banks are not only exposed to the three risks covered under Pillar 1; they are also exposed, for example, to securitization, concentration, liquidity, business, and residual credit risks. The Supervisory Review and Evaluation Process, which is the Euro-pean version of the capital review under Pillar II, requires banks to take a forward- looking and more comprehensive view on risk, which in turn should determine both capital and strategic planning. Supervisors review and evaluate the banks’ internal stress testing framework according to the principle of proportionality and make regular and comprehensive assessments of the pro-cesses, strategies, and systems that banks integrate into their Internal Capital Adequacy Assessment Process. Supervisors consider banks’ results to assess capital adequacy and, if necessary, require additional capital and liquidity buffers.

The Basel Committee on Banking Supervision and country authorities have issued guidance on the implementation of stress tests by banks. During the global financial crisis, the Basel Committee developed Principles for Sound Stress Testing Practices and Supervision (BCBS 2009) after reviewing supervisory authorities’ implementation of stress tests. These principles were mir-rored in guidelines developed by supervisory authorities (CEBS 2010). They state that stress scenarios should (1) reflect bank- specific risks, (2) take into account system- wide interactions, and (3) be flexible enough to adapt to changes in portfolio composition, the emergence of new risks, and specific risks related to businesses, entities, and products.38 These principles also anchor stress testing in the corporate governance and the riskmanagement culture of a bank. Senior management is required to (1) identify and communicate the level of risk appetite according to the bank’s business model, (2) identify relevant and plausible stress scenarios that are tested on a firm- wide level, and (3) ensure that results feed into the bank’s decision- making process, including strategic business planning and capital and liquidity planning.

However, the BCBS (2012) found that these guidelines were not followed systematically. Risk managers were not able to communicate the purpose, results, and implications of their assessments to the risk owners within the banks. In many cases, stress testing remained a very technical exercise, which was completed in a rather isolated and mechanical manner without be-ing sufficiently integrated with business lines. Consequently, the results were often interpreted as unrealistic, not credible, or just too technical. In many cases, these tests were performed separately for different business units, which prevented a compre-hensive, firm- wide perspective across different units and risks. Consequently, risks were largely underestimated. Moreover, precrisis stress scenarios were most often based on data that did not cover heavy downturn scenarios and the simultaneous re-alization of several risks. This, in turn, caused a systematic underestimation of the scope and relevance of various shocks during the global financial crisis. The Basel Committee further argued that banks failed to integrate guidelines on reputational risk and risks arising from off- balance- sheet vehicles, as well as the risks arising from highly leveraged counterparties and deficien-cies in risk- mitigating techniques. In 2017, the Basel Committee revised the 2009 stress testing principles, which are now stated at a higher level to enhance greater acceptance and applicability while preserving necessary flexibility to accommodate new developments (BCBS 2018).39

38 Also, scenarios should feature a wide range of alternatives, from optimistic forecasts (“baseline scenario”) to tail events challenging the bank’s business model (that is, “reverse stress tests”).

39 The Basel Committee published these principles following public consultation (BCBS 2018).


Appendix 2.2.Stress Testing of Nonbank Financial

Institutions: Insurance Companies and Financial Market Infrastructures

INSURANCE COMPANIESStress testing of the insurance sector is gaining growing acceptance as a riskmanagement tool, especially as developments in macroprudential surveillance have warranted greater focus on identifying systemic risk affecting the insurance sector. Insurers increasingly use it to assess and manage their risks. Insurance supervisors use it to assess the risks facing specific insurers and to identify possible vulnerabilities of the sector as a whole. Even though traditional insurance activities have not contributed to systemic risk during the financial crisis, there are some vulnerabilities from nontraditional, noninsurance activities (IAIS 2013), which could change the transmission channels of risks affecting the solvency and liquidity conditions of insurance companies under stress. If an insurer is significantly involved in activities that are closely linked to the broader financial sector— such as providing protection against credit exposures— then its failure could have systemic implications. Similarly, the failure of a large insurance company for which there is no quick substitution in the market, would also make a systemic case (Geneva Associa-tion 2010).

However, insurance stress testing has become an important element of stress testing within Financial Sector Assessment Programs (FSAPs) following the global financial crisis. Prior to that, it had typically played a secondary role relative to the analysis of the banking sector risks. Before the crisis, only 13 FSAPs (out of a total of almost 170) contained insurance stress tests (of which 11 were completed in advanced economies). This might be explained not only by the fact that insurers are con-sidered less systemically relevant in many jurisdictions but also by the unique conceptual challenges that emerge from the dif-ferent balance sheet structure of insurance companies and lack of a global solvency and valuation standards, which has resulted in a greater reliance on national supervisory frameworks for more resource- intensive stress testing in bottom- up approaches (Jobst, Sugimoto, and Broszeit 2013).

In FSAP insurance stress tests, efforts have been made to align the economic shocks to those used for the banking stress tests. However, insurers are also vulnerable to other types of risk that exceed the traditional parameters of bank stress testing and are difficult to capture in general macroeconomic scenarios. For example, natural events, such as floods, earthquakes, and windstorms, can be important to non- life insurers; and life insurers might be severely affected by a pandemic or by long- term improvements in mortality. The effects of stresses on the insurance sector are particularly difficult to test on a top- down basis because of the contract- level linkage between assets and liabilities for many life insurance products and the effects of insurer- specific reinsurance programs on the financial condition of non- life insurers. In addition, many insurance supervisors— even in developed markets— do not have the detailed data and models needed to perform such tests. It might be argued that top- down insurance stress testing at a meaningful level of granularity is impossible in most jurisdictions, at least until supervisory data and modeling capabilities have evolved, and that efforts should instead be made to improve the quality of bottom- up stress testing.

STRESS TESTING OF FINANCIAL MARKET INFRASTRUCTURES (FMI)40

Stress testing has also been applied to assessing the resilience of financial market infrastructures. In contrast to what is done for banks, it is not their balance sheet that is tested, but their proper functioning in case a risk materializes. This risk may be of an operational, credit, or liquidity nature. What matters is the immediate reaction of the FMI, the way it will finish the day and be able to operate the next few days following the shock. Both central banks, as FMI operators and overseers, and FMIs

40 FMIs refer to payment systems, central securities depositories, securities settlement systems, and central counterparties.



themselves conduct stress tests. In FSAPs, the IMF staff helps specify the stress testing requirements embedded in the Commit-tee on Payment and Settlement Systems (CPSS)/International Organization of Securities Commissions (IOSCO) standards and determines whether FMIs satisfy those requirements.

Stress Testing by Central Banks

Central banks play a crucial role in safeguarding the safety and soundness of payment systems. They also seek to ensure pay-ment systems operate in a practical and efficient manner for users and for the economy as a whole. In most cases, central banks operate the main national large- value payment system, which is the backbone of the entire financial market infrastructure. In addition, central banks are often in charge of overseeing core payment systems and other FMIs, such as central counterparties (CCPs) and securities settlement systems, such as central securities depositories (CSDs).

As operators and supervisors of the main payment system, some central banks use simulation tools to analyze the underlying payment flows and participant behavior in different scenarios— in addition to a standard- based qualitative approach. Opera-tional disruptions of the FMI itself (or a major participant) are often tested as well as the financial default of major participants. Incidents are simulated in order to identify recovery times, critical participants, contingency measures, and stop- sending limi-tations under different parameters, such as the concentration level, availability of liquidity, back- up procedures, reactions of nondefaulted banks, and structure of the money market.41 They can also help quantify the impact on liquidity by simulating the suspension of a participant’s outgoing payments, which results in liquidity accumulations for the failing participant and liquidity shortages for other participants and thereby disrupts settlements. Financial defaults of major participants are also tested to check whether the payment system will be able to handle them properly and to assess the impact on remaining partici-pants. Findings from simulation exercises could give rise to operational, organizational, or financial changes, such as the imple-mentation of new and/or more robust risk- mitigation facilities.42

Stress Testing by FMIs

In the wake of the financial crisis, many FMIs have developed their own stress testing approaches to manage liquidity and credit risks (BIS/IOSCO 2012).

Payment and Settlement Systems

Aside from central banks, several large private settlement systems have developed stress testing processes. The Clearing House Interbank Payment System, the US private sector large- value system, has instituted a program to ensure that participants un-derstand the consequences of a failure of one or more banks to honor their closing positions and to encourage participants to develop liquidity contingency plans. The Continuous Link Settlements Bank, which operates payment- versus- payment settle-ment services in 17 currencies, covering more than 60 percent of foreign exchange transactions worldwide, conducts a range of stress testing and back- testing scenarios to review the adequacy of its riskmanagement procedures. The simulations include, among other things, the adequacy of haircuts and the failure of the settlement member with the single largest funding obliga-tion in a single currency. Since the global financial crisis, the Continuous Link Settlements Bank has been working on more extreme scenarios, such as the failure of all settlement members of a given currency to complete their funding and the simulta-neous failure of all liquidity providers in the same currency.43 Some securities settlement systems, in particular international CSDs,44 which also offer securities loans and credit lines to participants, face potentially large credit and liquidity risks. Most of them conduct regular stress tests of their financial resources.

Central Clearing

Among FMIs, CCPs45 exhibit the highest concentration of liquidity and credit risks. Their core service is to become principal to every transaction that they clear, which implies that market participants no longer have credit exposures to their trading

41 The simulations can, for example, indicate if the system can complete settlement before the end of the day, as prescribed by CPSS standards, and allow defining the level of the contingency capacity needed.

42 The Bank of Finland has developed a payment system simulator ( BoF- PSS2), available to other central banks, which may customize it. The Bank of Fin-land also organizes regular seminars among central banks to share its practical experience in stress testing and define the business requirements of the next version of the simulator. Eurosystem overseers have decided to develop a tailor- made TARGET2-specific simulator based on BoF- PSS2 to run quantita-tive simulation- based stress tests on TARGET2.

43 For more detail on the Clearing House Interbank Payment System and the Continuous Link Settlements Bank’s stress test, see IMF 2010d.44 The two main international CSDs are Euroclear and Clearstream.45 The main international CCPs (in order of scale of operations) are CME Group, LCH, Clearnet, Eurex Clearing, Fixed Income Clearing Corporation,

Japan Government Bond Clearing Corporation, and Cassa di Compensazione e Garanzia.


counterparties, but only to the CCP. Therefore, CCPs concentrate credit risk and would face large liquidity needs if a partici-pant defaults because they need to fulfill the settlement obligations of the defaulting participant, potential losses when the cleared position or related collateral are liquidated, and the cash flows relating to possible hedge transactions.

Stress testing is key in managing the credit and liquidity risks of CCPs. Stress tests consider extreme but plausible market conditions and are typically framed in terms of the number of participant defaults a CCP can withstand. Current CPSS/IOSCO standards (BIS/IOSCO 2012) prescribe that CCPs should be able to withstand the default of the participant with the largest exposure, but some CCPs have chosen to be more stringent and test for the default of several major participants. CCPs’ models that calculate margin requirements, default fund contributions, collateral requirements, and other risk control mecha-nisms are expected to be subjected to rigorous and frequent stress tests that reflect their product mix and other riskmanagement choices. Key elements of stress tests are the assumed market conditions and default scenarios and the test frequency. A CCP should assume extreme market conditions (that is, price changes significantly larger than the prevailing levels of volatility), and evaluate the potential losses in individual participants’ positions. Other stress tests may consider the distribution of positions between the defaulting participants and their customers in evaluating potential losses. These should consider the resources of the potential defaulters that are available to a CCP (margins, clearing fund contributions, or other assets), as well as the CCP’s own resources, to provide perspective on the potential size of the losses and liquidity gaps of the CCP.

Since the financial crisis, the attention of regulators and supervisors has been drawn to CCPs’ stress testing, in particular when clearing over- the- counter (OTC) derivatives. Central clearing of OTC derivatives presents more challenges than clearing listed or cash- market products because of their complex risk characteristics. Following the G20 commitment to strengthen regulation of the OTC derivatives markets, improved rules took effect at the end of 2012. In some jurisdictions, this has in-cluded guidance on stress testing. For example, the European Commission adopted technical standards that specify the types of tests to be undertaken for different classes of financial instruments and portfolios, the involvement of clearing members or other parties in the tests, the frequency of tests, and the time horizon. In addition, CCPs’ supervision and oversight have gener-ally been strengthened since the crisis, for example by systematically analyzing CCPs’ internal riskmanagement models, in-cluding stress testing parameters.

Stress Testing by the IMF

The IMF often examines FMI stress testing arrangements in the context of FSAPs as well as Stability Modules (including re-views of standards and codes and reviews under the Offshore Financial Center Assessment Program). So far, the IMF has not conducted stress tests on FMIs. However, the IMF specifies the stress testing requirements embedded in the CPSS/IOSCO standards (BIS/IOSCO 2012) and checks whether those requirements are met by the respective FMIs.



Appendix 2.3.Prominent Crisis- Management

Stress Tests

Since the global financial crisis, stress tests have increasingly been used as instruments for crisis management. Their results feed into political decision- making processes, and in many cases determine the allocation of public support measures.

The most prominent examples for crisis management stress testing exercises were those conducted in the United States and Europe: (1) the US Federal Reserve’s Supervisory Capital Assessment Program (SCAP), which was completed in May 2009 (Board of Governors of the US Federal Reserve System 2009); (2) the EU system- wide stress tests performed by the Committee of European Banking Supervisors in 2010 (CEBS 2010) and the European Banking Authority (EBA) in 2011 (EBA 2011a, 2011b); and (3) the EBA’s 2011–2012 Capital Exercise (EBA 2012). In contrast, the US Federal Reserve’s Comprehensive Capi-tal Analysis and Review (Board of Governors of the US Federal Reserve System 2009), which evolved from the SCAP in 2011, the EU system- wide stress tests since 2014, which evolved into a biannual exercise (EBA 2014, 2016, 2017), and routine tests required under the Basel frameworks constitute typical supervisory stress tests.

KEY FEATURES OF CRISIS- MANAGEMENT STRESS TESTS Crisis- management stress testing exercises incorporate characteristics of both microprudential and macroprudential (or surveil-lance) stress tests. Like in supervisory stress tests, the individual institutions’ resilience to shocks is assessed on a bank- by- bank basis. The tests incorporate common macroeconomic and market- level shocks, and evaluate the banks’ performance relative to either existing (or future) regulatory requirements or alternative thresholds and definitions for capital ratios specifically de-signed for the particular exercise.46 Potential follow- up actions for banks that do not pass the test include private recapitaliza-tion and/or the acceptance of governmental support, mandatory restructuring, and changes in business lines/models.47

Crisis- management exercises are conducted for specific (crisis management) purposes and do not take place on a regular basis. The shock scenarios usually involve higher probability shocks that are more likely to materialize than the extreme- but- plausible tail risks typically tested in surveillance and supervisory tests. The exercise is comprehensive, covering both the bank-ing and trading books, and usually consider all off- balance sheet positions. Portfolios are broken down by geographic regions, industry sectors, and asset classes, which in turn implies that potential losses can differ across banks not only because of their different portfolio composition, but also reflecting differences in asset quality.

Crisis- management stress tests are designed as traditional balance- sheet- based solvency tests. The SCAP, EU system- wide stress testing exercises, and the EBA Capital Exercise focused on solvency risk in individual banks; liquidity risk was either not tested (SCAP) or assessed in a separate exercise (EBA 2011b).48 Spillover effects and default dependencies across institutions were, to some extent, indirectly considered through the design and structure of adverse scenarios, which typically take into ac-count several transmission channels. These exercises applied a form of static or constant balance sheet assumption to allow banks to shrink balance sheets or adapt business models during the forecasting period. Such approaches eliminate the scope for strategies to boost capital ratios by reducing risk- weighted assets. At the same time, this comes at the price of disregarding be-havioral response functions.

46 For instance, in the 2011 EU system- wide stress test, the EBA focused on banks’ core capital and considered a specific Core Tier 1 capital definition that was based on the Capital Requirements Directive II definition of Tier 1 capital net of deductions for participation in financial institutions, excluding hybrid instruments but including existing preference shares and existing governmental support measures. This definition should not be confused with Common Equity Tier 1 under Basel II (and implemented in the EU under the Capital Requirements Directive IV). In the 2009 SCAP, the US Federal Reserve applied a modified definition of Tier 1 capital, which excluded preferred stock, minority interest in subsidiaries, and less qualifying trust pre-ferred securities.

47 At the same time, crisismanagement stress tests do not limit their focus to assessing the health of individual banks but are equally concerned about the stability of the system as a whole. In order to evaluate the resilience of the banking system and the potential need for macroprudential or system- wide measures, aggregate indicators are taken into consideration.

48 The 2011 EU system- wide stress test, however, included a shock to banks’ funding costs (that is, funding liquidity) within the solvency stress testing framework. A traditional liquidity stress test assessing banks’ liquidity profiles was performed separately. The results of this assessment were not published.



Crisis- management stress tests combine elements of the bottom- up and top- down approaches.49 Based on a detailed meth-odology designed by the supervisor, banks examine the impact of one or more common macroeconomic scenarios on their portfolios. The results are checked for completeness, consistency, and plausibility by the supervisor. But the tests also include centralized components.50 The idea is to maximize the benefit of bottom- up approaches and, at the same time, to cross- check or validate the banks’ results through top- down elements.51

COMPARING THE CRISIS- MANAGEMENT STRESS TESTING EXERCISESThe three most prominent crisismanagement exercises had different purposes. The goal of the SCAP was to “[e]nsure adequate system capital to promote lending and restore investor confidence” (Board of Governors of the US Federal Reserve System 2009). The exercise was designed to “estimate losses, revenues, and reserve needs” for the large bank holding companies, and evaluated the size of governmental capital injections contingent on the banks’ performance in the stress test (Board of Gover-nors of the US Federal Reserve System 2009). The EU system- wide stress testing exercises aimed at examining the resilience of the banking system, identifying vulnerabilities, and informing policymakers about the current capacity of banks to absorb shocks and the banking system’s dependence on public support measures (EBA 2011a, 2011b). EBA’s Capital Exercise was spe-cifically designed to “create an exceptional and temporary capital buffer to address current market concerns over sovereign risk and other residual credit risk related to the current difficult market environment” (EBA 2012) but also informed a more objec-tive capital assessment of euro area banks as national authorities were preparing for the integrated supervision of systemically important institutions (via the Single Supervisory Mechanism).

The design and severity of macroeconomic stress scenarios depended on the objectives and the timing of each exercise. The adverse scenarios in both the SCAP and the EU system- wide stress tests involved a simultaneous realization of several risk fac-tors. EBA’s Capital Exercise, in contrast, was based on a EU system- wide baseline scenario provided by the European Commis-sion, with a focus on sovereign risks. Banks were required to conservatively assess the value of direct exposures to European Economic Area sovereigns held in the banking book (that is, loans and nontraded assets) at market prices as of September 2011. The adverse scenario of the 2011 (2010) EU system- wide stress test translated into a cumulative GDP shock of –4.1 (–3.1) per-centage points, compared with a –2.8 percentage point shock in the SCAP in 2009. Since the SCAP took place at the peak of the crisis, the assumed macroeconomic shock implied a deeper contraction compared with the European tests (that is, a cumu-lative two- year contraction in GDP of –2.7 percent under the SCAP, compared to a cumulative contraction of –0.2 [–0.4] per-cent in the 2011 [2010] EBA exercise).

Another key difference between these exercises was the organizational complexity of the EU system- wide tests. The EU ex-ercises involved 21 countries, 24 national supervisors and authorities, around a dozen European institutions, and 91 participat-ing banks (Appendix Figure 2.3.1). They had to deal with 19 different languages and seven currencies. None of these complexities applied to the SCAP.

The experience with these exercises was mixed. While both the SCAP and the EU system- wide stress testing shared (almost) identical goals, their impact was rather different:52

• The SCAP demonstrated convincingly that stress testing can be a powerful instrument for crisis management. The setup of the SCAP was credible, and backstop measures (including, crucially, government support) were clearly communicated and in place ahead of time. The US Federal Reserve argued that the test was “an important turning point in the finan-cial crisis” and that confidence improved as banks raised capital, mainly in the private markets. The results allowed markets to differentiate between banks based on their ability to withstand the shocks; after the publication of the re-sults, the correlation among major banks’ stock prices fell by more than 10 percent. Capital injections for banks that did not pass the tests were mandatory. Banks unable to raise capital in the private markets within a preset time period were asked to accept governmental capital support. Consequently, both goals of the SCAP were achieved— confidence in the US banking sector was restored and capital buffers were replenished to withstand further shocks to the system.

49 As the supervisory methodology constrains banks in assessing the impact of the scenarios, the EBA has coined the term “constrained bottom- up” tests.50 The 2009 SCAP basically followed a top- down approach and, at the same time, incorporated several decentralized components of bottom- up frame-

works, like the integration of banks’ own projections of losses, operating profits, and loan loss provisions under the given scenario. According to the de-tailed methodology designed by the supervisor, the CEBS/EBA tests asked banks to examine the impact of two scenarios on their portfolios. The results were checked and challenged internally by CEBS (2010) and within a multilateral review process, which was also flanked by top- down calculations by the European Systemic Risk Board in the 2011 exercise.

51 This is most crucial for preimpairment income, which serves as a first, and substantial, cushion against losses. Since banks have a better understanding of their business conditions, they are in a better position when discussing their bottom- up forecasts with the supervisor. While the US Federal Reserve ap-plied top- down stress tests to challenge the banks’ submissions, the EBA chose to cap net interest income at 2010 levels in its 2011 EU system- wide test.

52 While the SCAP exercise morphed into the annual Comprehensive Capital Analysis and Review (with the Dodd-Frank Annual Stress Testing as stress testing element) (Board of Governors of the US Federal Reserve System 2009, 2012), European authorities also have established system- wide stress testing as a regular macroprudential exercise [at two- year intervals] (EBA 2011a, 2011b, 2014, 2016, 2017).



• The 2011 EU system- wide stress test was in many aspects a successful exercise but did not fully achieve its goals. The exercise remedied several shortcomings of previous stress tests in 2010, and was praised for the risk coverage, tighter definition of capital components, quality assurance process, and detailed disclosure of exposures and results. The inclusion of stress on funding costs, sovereign exposures, and securitization positions expanded the test’s risk coverage substantially. While the tougher capital definition was generally welcomed, it was also criticized for being inconsistent with the Basel III definition of Common Equity Tier 1. The comprehensive publication of the methodology and results made the test an important exercise in disclosure, allowing analysts to replicate the findings and draw their own conclusions. How-ever, the whole exercise also contained several shortcomings. First, the scope and severity of the shock to sovereign ex-posures was deemed insufficient. While sovereign risk was, in principle, covered, the stress applied on sovereign exposures did not reflect the dislocations in European sovereign debt markets.53 Second, the adverse scenario was de-signed as a common scenario for all EU member states, but this meant that it did not adequately reflect the specific risk profiles of some national banking sectors, which was underscored by the failure of the Belgian bank Dexia in 2011 (which passed the stress test but become insolvent only a few weeks after the conclusion of the exercise). And third, backstops were seen by the markets as ambiguous, along with uncertainty over actions for banks ending up slightly above the hurdle rate, which considerably undermined attempts to restore investor confidence.

53 “The banking component can no longer be separated from sovereign and institutional developments. This is why Friday’s publication of stress tests results, while useful, is unlikely to be the game- changer it could have been two years ago (Véron 2011).”

Appendix Figure 2.3.1. Institutional Setup of EBA Stress Tests (2011)

EBABoard of Supervisors/Management Board

Quality AssuranceTask Group

Thematic Sub-Groups

EBA Committees

EBA Stress Test Team European Central Bank

European Commission

Euro Area

EU Committees

Participating Banks

Stress Test Expert Group(EBA, EC, ECB/ESRB, NCB, NSA)

NCBs and NSAs; MOFs

European SystemicRisk Board (ESRB)

Working Group onBackstop Measures

External Stakes(for example, IMF,

Governments)

Markets andBroader Public

Source: Authors.Note: EBA = European Banking Authority; EC = European Commission; ECB = European Central Bank. MOFs = Ministries of Finance; NCBs = national central banks; NSAs = national supervisory authorities.


Appendix 2.4.Frontiers of Stress Testing

This appendix presents some obvious conceptual challenges for stress testing and various analytical approaches that might be used to address them, including work being done at the IMF.

INTEGRATING SOVEREIGN RISK AND BANKING- SOVEREIGN FEEDBACKS IN STRESS TESTSConceptual and technical difficulties related to how to account for the impact of sovereign- bank links became evident in the EU system- wide stress testing. However, it is difficult to design stress testing models that account for banks’ impact on sover-eign risk and the feedback from sovereigns to banks through several transmission channels (Appendix Figure 2.4.1). The mark- to- market fall in the value of sovereign bonds held by banks reduces bank asset values, and distressed banks increase explicit or implicit) contingent liabilities of the public sector. Higher contingent liabilities might lead to rising sovereign spreads, which, in turn, can raise bank funding costs. If the sovereign is distressed enough, the value of official support (guarantees) to the bank is eroded. This can have knock- on effects on foreign banks and other sovereigns, as shown in Appendix Figure 2.4.1.

To include sovereign risk in stress testing, the key interlinked risk exposures between the government and financial sector should be analyzed in a comprehensive framework. A stylized framework starts with the economic, that is, risk- adjusted, bal-ance sheets of the financial sector, which can be treated as a portfolio and linked to the government’s balance sheet. In this specification, distressed financial institutions can lead to large government contingent liabilities, which in turn reduce govern-ment assets and lead to higher risk of default on sovereign debt. Dynamic macro- financial linkage models used for bank stress tests can also be linked to sovereign risk models, together with the feedback of banking risk to sovereigns via contingent liabili-ties and sovereign spreads affecting bank funding costs.

Sources: IMF 2010b; Jobst and Gray 2013; and authors.

Appendix Figure 2.4.1. Spillovers from Sovereigns to Banks and Banks to Sovereigns

1. Lower mark to market in valueof all government bonds held by local banks

2. Increase in bank funding costs3. Erosion in potential for official support/bailout

Increase in contingent liabilities

Increase in contingent liabilities

Lower mark to market in value ofall government bonds held by foreign banks

Sovereign

ForeignSovereign

Banks

ForeignBanks

Similar sovereigns comeunder pressure

Rise in counterpartycredit risk

Foreigneconomy

Domesticeconomy



ENHANCING ANALYSIS OF MACROECONOMIC AND BANKING SECTOR FEEDBACK AND CONTAGION BETWEEN FINANCIAL INSTITUTIONSA typical stress testing exercise uses macroeconomic scenarios to assess the impact on the banking/financial sector risk and capital adequacy without feedbacks to the macroeconomy. However, in many cases, distress in the banking/financial sector precedes the contraction of credit, leading to lower GDP growth. This feedback effect tends to be absent in stress tests. Models that contain such feedback channels would be useful, for example, a shock to the financial sector might be used to estimate the reduction in GDP growth and other factors, which would in turn have negative impacts on corporate and household borrow-ers, which would in turn increase credit risk on banks’ balance sheets.54 Dynamic factor models can help enhance the modeling of the feedback between the banks and the macroeconomy.

The global financial crisis demonstrated the potential for strong contagion among financial institutions. Traditional stress tests that use macroeconomic factor models to link to banking risk have a certain built- in correlation between banks, which comes from their common correlations to the macroeconomic factors. However, this does not capture correlations, dependen-cies, and feedback between the institutions. Contagion effects can be modeled with networks, joint default probabilities or expected losses, and a variety of other models that have been developed. Enhanced stress testing can include some of the fea-tures of these systemic risk models to improve the analysis of interdependence and joint risk.

TECHNIQUES TO IMPROVE MODELING OF NONLINEARITIES IN CHANGES IN BANK ASSETS, CAPITAL, CREDIT RISK, AND FUNDING COSTThe risk- adjusted balance sheet can be helpful in illustrating the trade-offs between the nonlinear impact of changes in bank’s assets and (risky) debt on credit/funding spreads and capital. The fundamental conceptual framework of the risk- adjusted bal-ance sheet comes from the contingent claims approach (Merton 1973) and risk- neutral valuation (Cox and Ross 1976). A bank’s liabilities (equity and [risky] debt) are claims on underlying bank assets that are uncertain over this time horizon, and the degree of uncertainty (that is, volatility) affects the risk premiums and values of equity and debt liabilities. There are differ-ent ways to construct the risk- adjusted balance sheet. One can use the estimated loan portfolio loss distribution and other components of the bank’s balance sheet. Using risk- neutral valuation, the probability distribution of the bank’s risky loans and distribution of assets (over a specific time horizon) can be estimated. This asset distribution is then combined with the prom-ised payments on debt and deposits to construct a risk- adjusted ( contingent- claims- analysis- type) balance sheet. This technique does not rely on market prices. A second method estimates the market- implied asset and asset volatility of a bank from the observed market value and volatility of the bank’s equity and the book value of its debt and deposits. Comparing the results of the two methods can provide insights into the dynamics and differences between the “fundamental” loan portfolio loss ap-proach and the market-implied view (which will vary between calm and stress periods).

This modeling approach can also be used to estimate the increase in funding cost dependent on the risk- free interest rate, the bank’s credit spread, market risk appetite, and the impact of the government’s (implicit and explicit) guarantees. In addi-tion, the risk- adjusted balance sheet of a bank already incorporates the dynamic changes between assets and market capital.55 Changes in assets lead to changes in market equity and changes in expected losses to bank creditors or guarantors. The magni-tude of these losses depends on the level of distress of the bank. Thus, the change in bank market capitalization due to the de-cline in implied asset values is analogous to an aggregate risk- weighted asset adjustment factor in the Basel III framework, and can be interpreted as a “ quasi- risk- weighted assets” dynamic adjustment factor.

54 See Krznar and Matheson 2017.55 More work is needed on capturing the nonlinearities between changes in bank assets and capital. In balance- sheet- based stress testing models, fixed cor-

relations between exposures and static risk weights can lead to underestimation of the capital shortfall under stress. This can be improved by using time- varying correlation between exposures when estimating portfolio loss distributions.



———. 2011. Global Systemically Important Banks: Assessment Methodology and the Additional Loss Absorbency Requirement. Basel: Bank for International Settlements. https://www.bis.org /publ/bcbs207.htm.

———. 2012. Peer Review of Supervisory Authorities’ Implementa-tion of Stress Testing Principles. Basel: Bank for International Settlements. https://www.bis.org/publ/bcbs218.htm.

———. 2017a. Finalizing Post- Crisis Reforms. Basel: Bank for Inter-national Settlements. https://www.bis.org/bcbs/publ/d424.htm.

———. 2018. Stress Testing Principles. Basel: Bank for Interna-tional Settlements. https://www.bis.org/bcbs/publ/d450.htm.

Board of Governors of the US Federal Reserve System. 2009. The Supervisory Capital Assessment Program: Design and Implemen-tation. Washington, DC: Federal Reserve.

———. 2012. United States Comprehensive Capital Analysis and Review 2012: Methodology and Results for Stress Scenario Projec-tions. Washington, DC: Federal Reserve.

Borio, Claudio, and Mathias Drehann, 2009. “Towards an Opera-tional Framework for Financial Stability: ‘Fuzzy’ Measurement and Its Consequences.” BIS Working Paper 284, Bank for In-ternational Settlements, Basel, Switzerland. https://www.bis .org/publ/work284.htm.

Borio, Claudio, Mathias Drehmann, and Kostas Tsatsaronis. 2014. “ Stress- testing Macro Stress Testing: Does It Live up to Expectations?” Journal of Financial Stability 12 (1): 3–15.

Brunnermeier, Markus K., and Lasse Heje Pedersen. 2009. “Mar-ket Liquidity and Funding Liquidity.” Review of Financial Studies 22 (6): 2201–38.

Čihák, Martin. 2007. “Introduction to Applied Stress Testing.” IMF Working Paper 07/59, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP / I s sue s /2016/12/31/ Int roduc t ion-to -Appl ied-St re s s -Testing-20222.

Čihák, Martin, Sonia Muñoz, and Ryan Scuzzarella. 2011. “The Bright and the Dark Side of Cross- Border Banking Linkages.” IMF Working Paper 10/105, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ The- Bright- and- the- Dark- Side- of -Cross-Border-Banking-Linkages-25147.

Committee of European Banking Supervisors (CEBS). 2010. CEBS Guidelines on Stress Testing. London: Committee of European Banking Supervisors. https://www.eba.europa.eu/ regulation- and - policy/ supervisory- review- and- evaluation- srep- and- pillar- 2 /revised-guidelines-on-stress-testing.

Cox, John C., and Stephen A. Ross. 1976. “The Valuation of Op-tions for Alternative Stochastic Processes.” Journal of Financial Economics 3 ( January– March): 144–66.

Ennis, Huberto M., Helen Fessenden, and John R. Walter, 2016. “Do Net Interest Margins and Interest Rates Move Together?” US Federal Reserve Bank of Richmond Economic Brief EB16-05, Federal Reserve, Richmond, May. https://www.richmond fed .net/publications/research/economic_brief/2016/eb_16-05.

Espinosa- Vega, Marco A., and Juan Sole. 2010. “Cross Border Fi-nancial Surveillance: A Network Perspective.” IMF Working Paper 10/105, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 / Cro s s - B orde r- F i n a nc i a l - Su r ve i l l a nc e -A-Ne t work -Perspective-23788.

Espinoza, Raphael, and Miguel Segoviano. 2011. “Probabilities of Default and the Market Price of Risk in a Distressed Econ-omy.” IMF Working Paper 11/75, International Monetary

REFERENCESAcharya, Viral V., Lasse H. Pedersen, Thomas Philippon, and

Matthew P. Richardson. 2010. “Measuring Systemic Risk.” In Regulating Wall Street: The Dodd‐Frank Act and the New Archi-tecture of Global Finance, edited by Acharya, Viral V., Thomas F. Cooley, Matthew P. Richardson, and Ingo Walter. Hoboken: John Wiley & Sons.

Adrian, Tobias, and Markus Brunnermeier. 2016. “CoVaR.” Amer-ican Economic Review 106 (7): 1705-41.

Aikman, David, Piergiorgio Alessandri, Bruno Eklund, Prasanna Gai, Sujit Kapadia, Elizabeth Martin, Nada Mora, Gabriel Sterne, and Matthew Willison. 2009. “Funding Liquidity Risk in a Quantitative Model of Systemic Liquidity.” Working Paper 372, Bank of England, London. https://www.bankofengland .co.uk/ working- paper/2009/ fundingliquidity- risk- in- a -quantitative-model-of-systemic-stability.

Aït- Sahalia, Yacine, Jochen Andritzky, Andreas A. Jobst, Sylwia Nowak, and Natalia Tamirisa 2012. “How to Stop a Herd of Running Bears? Market Response to Policy Initiatives during the Global Financial Crisis.” Journal of International Economics 87 (1): 162–77.

Allen, Franklin, and Gale Douglas. 2000. “Financial Contagion.” Journal of Political Economy 108 (1): 1–33.

Avesani, Renzo, Kexue Liu, Alin Mirestean, and Jean Salvati. 2008. “Review and Implementation of Credit Risk Models of the Financial Sector Assessment Program.” IMF Working Pa-per 06/134, International Monetary Fund, Washington, DC. https://w w w.imf.org/en/Publicat ions/WP/Issues/2016 /12/31/ Review- and- Implementation- of- Credit- Risk- Models - of- the-Financial-Sector-Assessment-Program-19111.

Bank for International Settlements (BIS) and International Orga-nization of Securities Commissions (IOSC). 2004. Committee on Payment and Settlement Systems Technical Committee of the International Organization of Securities Commissions: Recom-mendations for Central Counterparties. Basel: Bank for Interna-tional Settlements. https://www.bis.org/cpmi/publ/d64.htm.

———. 2012. Committee on Payment and Settlement Systems Tech-nical Committee of the International Organization of Securities Commissions: Principles for Financial Market Infrastructures. Ba-sel: Bank for International Settlements. https://www.bis.org /cpmi/publ/d101.htm.

Barnhill, Theodore, Panagiotis Papapanagiotou, and Liliana Schumacher. 2002. “Measuring Integrated Credit and Market Risks in Bank Portfolios: An Application to a Set of Hypotheti-cal Banks Operating in South Africa.” Journal of Financial Markets, Institutions and Instruments 11 (5): 401–43.

Barnhill, Theodore, and Liliana Schumacher. 2011. “Modeling Correlated Systemic Liquidity and Solvency Risks in a Financial Environment with Incomplete Information.” IMF Working Pa-per 11/263, International Monetary Fund, Washington, DC. https://w w w.imf.org/en/Publicat ions/WP/Issues/2016 /12/31/ Modeling- Correlated- Systemic- Liquidity- and- Solvency - Risks- in-a-Financial-Environment-with-25356.

Basel Committee on Banking Supervision (BCBS). 2004. Princi-ples for the Management and Supervision of Interest Rate Risk. Basel: Bank for International Settlements. https://www.bis.org /publ/bcbs108.htm.

———. 2009. Principles for Sound Stress Testing Practices and Su-pervision. Basel: Bank for International Settlements. https://www.bis.org/publ/bcbs147.htm.


https://www.bis.org/publ/bcbs207.htm.



https://www.bis.org/bcbs/publ/d424.htm.


https://www.bis.org/publ/work284.htm






https://www.eba.europa.eu/

https://www.richmond

http://fed.net/publications/research/economic_brief/2016/eb_

http://fed.net/publications/research/economic_brief/2016/eb_



https://www.bankofengland.co.uk/




https://www.bis.org/cpmi/publ/d64.htm.

https://www.bis.org/cpmi/publ/d101.htm

https://www.bis.org/cpmi/publ/d101.htm





https://www.bis.org/publ/bcbs147.htm











.bankofengland.co.uk /f inancia l-stabi l it y-paper/2007 /a-new-approach-to-assessing-risks-to-financial-stability.

Hardy, Daniel, and Christian Schmieder. 2013. “Rules of Thumb for Solvency Stress Tests.” IMF Working Paper 13/232, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publi cations/WP/Issues/2016/12/31/Rules-of-Thumb-for-Bank -Solvency-Stress-Testing-41047.

Independent Evaluation Office (IEO) of the International Mone-tary Fund (IMF). 2006. Report on the Evaluation of the Finan-cial Sector Assessment Program. Washington, DC: International Monetary Fund. https://www.imf.org/External/NP/ieo/2006 /fsap/eng/.

International Association of Insurance Supervisors (IAIS). 2013. Global Systemically Important Insurers: Initial Assessment Meth-odology. Basel: Bank for International Settlements. https://www.iaisweb.org/page/supervisory-material/financial-stability -and-macroprudential-policy-and-surveillance//file/34257 /final-initial-assessment-methodology-18-july-2013.

International Monetary Fund (IMF). 2006. Report on the Evalua-tion of the Financial Sector Assessment Program. Washington, DC: International Monetary Fund. https://www.imf.org/en /Publications/Independent-Eva luation-Off ice-Reports /Issues/2016/12/31/IEO-Report-on-the-Evaluation-of-the -Financial-Sector-Assessment-Program-18711.

———. 2010a. “United States of America: Stress Testing—Tech-nical Note.” IMF Country Report 10/244, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /United-States-Publication-of-Financial-Sector-Assessment -Program-Documentation-Technical-24101.

———. 2010b. Global Financial Stability Report: Meeting New Challenges to Stability and Building a Safer System. Washington, DC, April. https://www.imf.org/en/Publications/GFSR/Issues /2016/12/31/Meeting-New-Challenges-to-Stability-and -Building-a-Safer-System.

———. 2010c. Global Financial Stability Report: Sovereigns, Fund-ing, and Systemic Liquidity, Chapter 1. Washington, DC, October. https://www.imf.org/en/Publications/GFSR/Issues /2016/12/31/Sovereigns-Funding-and-Systemic-Liquidity.

———. 2010d. Integrating Stability Assessments under the Finan-cial Sector Assessment Program into Article IV Surveillance: Back-ground Material. Washington, DC. https://www.imf.org/en /Publications/Policy-Papers/Issues/2016/12/31/Integrating -Stability-Assessments-Under-the-Financial-Sector-Assessment -Program-into-PP4478.

———. 2010e. “United States of America: Selected Issues on Li-quidity Risk Management in Fedwire Funds and Private Sector Payment—Technical Note.” IMF Country Report 10/122, Washington, DC. https://www.imf.org/en/Publications/CR / I s s u e s /2 016 /12 /31/ Un i t e d - S t a t e s -P u b l i c a t i on - o f -Financia l-Sector-Assessment-Program-Documentation -Technical-23864.

———. 2011a. Global Financial Stability Report: Durable Finan-cial Stability—Getting There from Here, Chapter 2. Washing-ton, DC, April. https://www.imf.org/en/Publications/GFSR /Issues/2016/12/31/Durable-Financial-Stability-Getting -There-from-Here.

———. 2011b. Global Financial Stability Report: Grappling with Crisis Legacies, Chapter 3. Washington, DC, September. https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31 /Grappling-with-Crisis-Legacies.

———. 2011c. “Luxembourg: Financial System StabilityAssess-ment—Update.” IMF Country Report 11/148, Washing-ton, DC. https://www.imf.org/en/Publications/CR/Issues/2016

Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/ Probabilities- of- Default- and- the - Market- Price- of- Risk-in-a-Distressed-Economy-24774.

European Banking Authority (EBA). 2011a. EU- Wide Stress Test-ing. Aggregate Report, London, http://www.eba.europa.eu / risk- analysis- and-data/eu-wide-stress-testing/2011/results.

———. 2011b. 2011 EU- Wide Stress Test: Methodological Note—Version 1.1. London, March 18.http://www.eba.europa.eu/-/ the - eba- publishes- deta i ls- of- it s-stress-test-scenarios-and -methodology.

———. 2012. Final Report on the Implementation of Capital Plans Following the EBA’s 2011 Recommendation on the Creation of Temporary Capital Buffers to Restore Market Confidence. Lon-don, July 11. https://www.eba.europa.eu/ risk- analysis- and- data /eu-capital-exercise/final-results.

———. 2014. EU-Wide Stress Test: Methodological Note. London, April 29. https://www.eba.europa.eu/-/eba-publishes-common -methodology-and-scenario-for-2014-eu-banks-stress-test.

———. 2016. EU- Wide Stress Test: Methodological Note. London, February 24. https://www.eba.europa.eu/-/eba- launches-2016 -eu-wide-stress-test-exercise.

———. 2017. EU-Wide Stress Test: Methodological Note. Lon- don, November 17. https://www.eba.europa.eu/-/eba-publishes -methodology-for-the-2018-eu-wide-stress-test.

European Central Bank (ECB). 2008. EU Banks’ Liquidity Stress Tests and Contingency Funding Plans. Banking Supervision Com-mittee (BSC). Frankfurt: European Central Bank.

Geneva Association. 2010. Systemic Risk in Insurance—An Analysis of Insurance and Financial Stability. Special Report of the Geneva Association Systemic Risk Working Group. Geneva: Interna-tional Association for the Study of Insurance Economics. https://www.genevaassociation.org/research-topics/financial-stability -and-regulation/systemic-risk-insurance-analysis-insurance -and.

Gray, Dale F., Robert C. Merton, and Zvi Bodie. 2008. “A New Framework for Measuring and Managing Macro-Financial Risk and Financial Stability.” Working Paper 09-015, Harvard Business School, Cambridge, MA. https://hbswk.hbs.edu/item /new-framework-for-measuring-and-managing-macro financial-risk-and-financial-stability.

Gray, Dale F., and Andreas A. Jobst. 2010. “Modelling Systemic Financial Sector and Sovereign Risk.” Sveriges Riksbank Eco-nomic Review 2: 68–106.

Greenlaw, David, Anil Kashyap, Kermit Schoenholtz, and Hyun Song Shin. 2012. “Stressed Out: Macroprudential Principles for Stress Testing.” Chicago Booth Research Paper 12-08, Uni-versity of Chicago, Chicago, IL.

Hannoun, Herve. 2011. “Sovereign Risk in Bank Regulation and Supervision: Where Do We Stand?” Speech by Herve Han-noun presented at High-Level Meeting for the Middle East and North Africa Region jointly organized by the Arab Monetary Fund and the Financial Stability Institute, Abu Dhabi, UAE, October 26. https://www.bis.org/speeches/sp111026.htm.

Haldane, Andrew. 2009a. “Why Banks Failed the Stress Test.” Speech at Marcus-Evans Conference on Stress Testing, Lon-don, February 9–10. https://www.bankofengland.co.uk/speech /2009/why-banks-failed-the-stress-test.

———. 2009b. “Small Lessons from a Big Crisis.” Remarks at the US Federal Reserve Bank of Chicago 45th Annual Conference, “Reforming Financial Regulation,” May 8.

———, Simon Hall, and Silvia Pezzini. 2007. “A New Approach to Assessing Financial Stability.” Bank of England Financial Stability Paper 2, Bank of England, London. https://www


http://www.bankofengland.co.uk/financial-stability-paper/2007

https://www.imf.org/en/Publi

https://www.imf.org/External/NP/ieo/2006/fsap/eng/

https://www.imf.org/External/NP/ieo/2006/fsap/eng/

https://www.iaisweb.org/page/supervisory-material/financial-stability-and-macroprudential-policy-and-surveillance//file/34257/final-initial-assessment-methodology-18-july-2013




https://www.imf.org/en/Publications/Independent-Evaluation-Office-Reports/Issues/2016/12/31/IEO-Report-on-the-Evaluation-of-the-Financial-Sector-Assessment-Program-18711.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-States-Publication-of-Financial-Sector-Assessment-Program-Documentation-Technical-24101.



https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Meeting-New-Challenges-to-Stability-and-Building-a-Safer-System.



https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Sovereigns-Funding-and-Systemic-Liquidity.

https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Sovereigns-Funding-and-Systemic-Liquidity.

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Integrating-Stability-Assessments-Under-the-Financial-Sector-Assessment-Program-into-PP4478.








https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Durable-Financial-Stability-Getting-There-from-Here.



https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Grappling-with-Crisis-Legacies.



https://www.imf.org/en/Publications/CR/Issues/2016



http://www.eba.europa.eu/

http://www.eba.europa.eu/-/


https://www.eba.europa.eu/-/eba-publishes-common-methodology-and-scenario-for-2014-eu-banks-stress-test.


https://www.eba.europa.eu/-/eba-launches-2016-eu-wide-stress-test-exercise.

https://www.eba.europa.eu/-/eba-launches-2016-eu-wide-stress-test-exercise.

https://www.eba.europa.eu/-/eba-publishes-methodology-for-the-2018-eu-wide-stress-test

https://www.eba.europa.eu/-/eba-publishes-methodology-for-the-2018-eu-wide-stress-test

https://www.genevaassociation.org/research-topics/financial-stability-and-regulation/systemic-risk-insurance-analysis-insurance-and




https://hbswk.hbs.edu/item/new-framework-for-measuring-and-managing-macro

https://hbswk.hbs.edu/item/new-framework-for-measuring-and-managing-macro

https://www.bis.org/speeches/sp111026.htm

https://www.bankofengland.co.uk/speech/2009/why-banks-failed-the-stress-test.

https://www.bankofengland.co.uk/speech/2009/why-banks-failed-the-stress-test.








http://www.bankofengland.co.uk/financial-stability-paper/2007


———. 2017. Global Financial Stability Report: Getting the Policy Mix Right, Chapter 2. Washington, DC, April. https://www . i m f .org /en /Publ ic a t ion s /GFSR / I s sue s /2017/03/30 /global-financial-stability-report-april-2017.

International Monetary Fund (IMF), Bank for International Set-tlements (BIS), and Financial Stability Board (FSB). 2009. Guidance to Assess the Systemic Importance of Financial Institu-tions, Markets, and Instruments: Initial Considerations. Report to the G20 Finance Ministers and Central Bank Governors. http://www.fsb.org/2009/11/r_091107c/.

Ishikawa, Atsushi, Kichiro Kamada, Yoshiyuki Kurachi, Kentaro Nasu, and Yuki Teranishi. 2012. “Introduction to the Financial Macroeconometric Model.” Bank of Japan Working Paper 12-E-1, Bank of Japan, Tokyo. https://www.boj.or.jp/en/research /wps_rev/wps_2012/wp12e01.htm/.

———. 2013. “Guidelines for the Bottom-Up Solvency Stress Test—Banking.” In “Belgium: Stress Testing the Banking and Insurance Sectors—Technical Note.” IMF Country Report 13/137, International Monetary Fund, Washington, DC. https://w w w.imf.org /en/Publ icat ions/CR /Issues/2016/12/31 /Belgium-Technical-Note-on-Stress-Testing-the-Banking-and -Insurance-Sectors-40573.

Jobst, Andreas A. 2014. “Measuring Systemic Risk-Adjusted Li-quidity (SRL)—A Model Approach.” Journal of Banking and Finance 44: 270–87.

Jobst, Andreas A., and Dale F. Gray. 2013. “Systemic Contingent Claims Analysis—Estimating Market-Implied Systemic Risk.” IMF Working Paper 13/54, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP / I s s u e s /2 016 /12 /31/Sy s t e m i c - C ont i n g e nt- C l a i m s -Analysis-Estimating-Market-Implied-Systemic-Risk-40356.

Jobst, Andreas A., Li Lian Ong, and Christian Schmieder. 2013.“A Framework for Macroprudential Bank Solvency Stress Testing: Application to S-25 and Other G-20 Country FSAPs.” IMF Working Paper 13/68, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/A-Framework-for-Macroprudential-Bank-Solvency -Stress-Testing-Application-to-S-25-and-Other-G-40390.

———. 2017.“Macroprudential Liquidity Stress Testing in FSAPs for Systemically Important Financial Systems.” IMF Working Paper 17/102, International Monetary Fund, Washington, DC. https://w w w.imf.org/en/Publicat ions/W P/Issues/2017/05/01/Macroprudential-Liquidity-Stress-Testing-in-FSAPs-for-Systemi-cally-Important-Financial-44873.

Jobst, Andreas A., Nobuyasu Sugimoto, and Timo Broszeit. 2014.“Macroprudential Solvency Stress Testing of the Insurance Sector.” IMF Working Paper 14/133, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/Macroprudential-Solvency-Stress -Testing-of-the-Insurance-Sector-41776.

J.P. Morgan. 1995. Creditmetrics. New York: J.P. Morgan & Co., Inc.

Krznar, Ivo, and Troy Matheson. 2017. “Towards Macroprudential Stress Testing: Incorporating Macro-Feedback Effects.” IMF Working Paper 17/149, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2017 /06/30/Towards-Macroprudential-Stress-Testing-Incorporating -Macro-Feedback-Effects-44955.

Le Leslé, Vanessa, and Sofiya Avramova. 2012. “Revising Risk-Weighted Assets: Why Do RWAs Differ across Countries and What Can Be Done About It?” IMF Working Paper 12/90, Inter-national Monetary Fund, Washington, DC. https://www.imf.org

/12/31/Luxembourg-Financial-System-Stability-Assessment -Update-24995.

———. 2011d. “United Kingdom: Stress Testing the Banking Sector—Technical Note.” IMF Country Report 11/227, Wash-ington, DC. https://www.imf.org/en/Publications/CR/Issues /2016/12/31/United-Kingdom-Stress-Testing-the-Banking -Sector-Technical-Note-25119.

———. 2011e. “Sweden: Stress Testing of the Banking Sector—Technical Note.” IMF Country Report 11/288, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12 /31/Sweden-Financial-Sector-Assessment-Program-Update -Technical-Note-on-Stress-Testing-of-the-25243.

———. 2011f. “Germany: Stress Testing—Technical Note.” IMF Country Report 11/371, Washington, DC. https://www.imf.org /en / P ub l i c a t ion s /C R / I s sue s /2 016/12 /31/G er m a ny -Technical-Note-on-Stress-Testing-25461.

———. 2012a. “Greece: Request for Extended Arrangement un-der the Extended Fund Facility—Staff Report.” IMF Country Report 12/57, Washington, DC. https://www.imf.org/en /Publications/CR/Issues/2016/12/31/Greece-Request-for -Extended-Arrangement-Under-the-Extended-Fund-Facility -Staff-Report-Staff-25781.

———. 2012b. “Macro-Financial Stress Testing—Principles and Practices.” IMF Policy Paper, Washington, DC. https://www .imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Macrofinancial-Stress-Testing-Principles-and-Practices-PP4702.

———. 2012c. “Macro-Financial Stress Testing: Principles and Practices—Background Material.” IMF Policy Paper, Wash-ington, DC. https://www.imf.org/en/Publications/Policy-Papers /Issues/2016/12/31/Macrofinancial-Stress-Testing-Principles -and-Practices-Background-Material-PP4703.

———. 2012d. “Enhancing Surveillance: Interconnectedness and Clusters—Background Paper.” IMF Policy Paper, Washing-ton, DC. https://www.imf.org/en/Publications/Policy-Papers / I s s u e s / 2 0 1 6 / 1 2 / 3 1 / E n h a n c i n g - S u r v e i l l a n c e -Interconnectedness-and-Clusters-Background-Paper-PP4721.

———. 2012e. “Israel: Stress Testing of the Banking, Insurance and Pension Sectors—Technical Note.” IMF Country Report 12/88, Washington, DC. https://www.imf.org/en/Publica t ions/CR /Issues/2016/12/31/Israel-Technica l-Note-on -Stress-Test-of-the-Banking-Insurance-and-Pension-Sectors -25850.

———. 2013a. “Mandatory Financial Stability Assessments under the Financial Sector Assessment Program–Update.” IMF Pol-icy Paper, Washington, DC. https://www.imf.org/en/Publica t ion s / Pol i c y-Paper s / I s sue s /2016/12/31/Ma nd ator y -Financial-Stability-Assessments-Under-the-Financial-Sector -Assessment-Program-PP4838.

———. 2013b. Global Financial Stability Report: Old Risks, New Challenges, Chapter 3. Washington, DC, April. https://www .imf.org/en/Publications/GFSR/Issues/2016/12/31/Old -Risks-New-Challenges.

———. 2014a. “IMF Executive Board Reviews Mandatory Finan-cial Stability Assessments under the Financial Sector Assess-ment Program.” Press Release No. 14/08 (January 13), Washington, DC. https://www.imf.org/en/News/Articles/2015 /09/14/01/49/pr1408.

———. 2014b. “People’s Republic of China–Hong Kong Special Administrative Region: Stress Testing the Banking Sector—Technical Note.” IMF Country Report 14/210, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ Peoples-Republic-of-ChinaHong-Kong-Special-Administra-tive-Region-Financial-Sector-Assessment-41755.





http://www.fsb.org/2009/11/r_091107c/

https://www.boj.or.jp/en/research/wps_rev/wps_2012/wp12e01.htm/.

https://www.boj.or.jp/en/research/wps_rev/wps_2012/wp12e01.htm/.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Belgium-Technical-Note-on-Stress-Testing-the-Banking-and-Insurance-Sectors-40573




https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Systemic-Contingent-Claims-Analysis-Estimating-Market-Implied-Systemic-Risk-40356



https://www.imf.org/en/Publications/WP/Issues/2016/12/31/A-Framework-for-Macroprudential-Bank-Solvency-Stress-Testing-Application-to-S-25-and-Other-G-40390.



https://www.imf.org/en/Publications/WP/Issues/2017/05/01/Macroprudential-Liquidity-Stress-Testing-in-FSAPs-for-Systemi-cally-Important-Financial-44873




https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Macroprudential-Solvency-Stress-Testing-of-the-Insurance-Sector-41776



https://www.imf.org/en/Publications/WP/Issues/2017/06/30/Towards-Macroprudential-Stress-Testing-Incorporating-Macro-Feedback-Effects-44955



https://www.imf.org/12/31/Luxembourg-Financial-System-Stability-Assessment-Update-24995.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-Kingdom-Stress-Testing-the-Banking-Sector-Technical-Note-25119.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Sweden-Financial-Sector-Assessment-Program-Update-Technical-Note-on-Stress-Testing-of-the-25243.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Germany-Technical-Note-on-Stress-Testing-25461.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Greece-Request-for-Extended-Arrangement-Under-the-Extended-Fund-Facility-Staff-Report-Staff-25781.




https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Macrofinancial-Stress-Testing-Principles-and-Practices-PP4702.



https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Macrofinancial-Stress-Testing-Principles-and-Practices-Background-Material-PP4703.



https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Enhancing-Surveillance-Interconnectedness-and-Clusters-Background-Paper-PP4721.



https://www.imf.org/en/Publica


https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Old-Risks-New-Challenges.



https://www.imf.org/en/News/Articles/2015/09/14/01/49/pr1408.

https://www.imf.org/en/News/Articles/2015/09/14/01/49/pr1408.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Peoples-Republic-of-ChinaHong-Kong-Special-Administra-tive-Region-Financial-Sector-Assessment-41755










Schmieder, Christian, Heiko Hesse, Benjamin Neudorfer, Claus Puhr, and Stefan W. Schmitz. 2012. “Next Generation System-Wide Liquidity Stress Testing.” IMF Working Paper 12/03, In-ternational Monetary Fund, Washington, DC. https://www .imf.org/external/pubs/cat/longres.aspx?sk=25509.0.

Segoviano, Miguel, and Charles Goodhart. 2009. “Banking Sta-bility Measures.” IMF Working Paper 09/04, International Monetary Fund, Washington, DC. https://www.imf.org/en /Publications/WP/Issues/2016/12/31/Banking-Stability -Measures-22554.

Segoviano, Miguel. 2006, “The Consistent Information Multivariate Density Optimizing Methodology.” Financial Markets Group, London School of Economics, Discussion Paper 557, London School of Economics, London. http://eprints.lse.ac.uk/24511/.

Taleb, Nassim N. 2004. Fooled by Randomness. New York: Thom-son/Texere.

Tressel, Thierry. 2010. “Financial Contagion through Bank Dele-veraging: Stylized Facts and Simulations Applied to the Finan-cial Crisis.” IMF Working Paper 10/236, International Monetary Fund, Washington, DC. https://www.imf.org/en /Publications/WP/Issues/2016/12/31/Financial-Contagion -Through-Bank-Deleveraging-Stylized-Facts-and-Simulations -Applied-to-the-24285.

Véron, Nicolas. 2011. “Stress Tests Fail to Rescue Europe’s Banks.” Blog Post, Bruegel, Brussels. http://bruegel.org/2011/07/stress-tests -fail-to-rescue-europes-banks/.

Wong, Eric, and Cho-Hoi Hui. 2009. “A Liquidity Risk Stress Testing Framework with Interaction between Market and Credit Risks.” Working Paper 06/2009, Hong Kong Monetary Authority, Hong Kong SAR. https://www.hkma.gov.hk/eng /publications-and-research/research/working-papers/2009/.

/en /Publ ic at ions / W P/I s sue s /2016/12/31/Rev i s it ing -Risk-Weighted-Assets-25807.

Matz, Leonard, and Peter Neu. 2006. Liquidity Risk Measurement and Management: A Practitioner’s Guide to Global Best Practices. Hoboken: John Wiley & Sons.

Merton, Robert C. 1973. “Theory of Rational Option Pricing.” Bell Journal of Economics and Management Science 4 (Spring): 141–83.

Office of Financial Research (OFR). 2012. 2012 Annual Report. US Department of the Treasury, Washington DC. https://www .financialresearch.gov/annual-reports/2012-annual-report/.

Ong, Li Lian, and Martin Čihák. 2010, “Of Runes and Sagas: Per-spectives on Liquidity Stress Testing Using an Iceland Exam-ple.” IMF Working Paper 10/156, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Of-Runes-and-Sagas-Perspectives -on-Liquidity-Stress-Testing-Using-an-Iceland-Example-24019.

Ong, Li Lian, and Ceyla Pazarbasioglu. 2014. “Credibility and Crisis Stress Testing.” International Journal of Financial Studies 2: 15–81. http://www.mdpi.com/2227-7072/2/1/15.

Reserve Bank of India. 2011. Financial Stability Report No. 4. Mumbai, Reserve Bank of India. https://rbi.org.in/Scripts/BS _PressReleaseDisplay.aspx?prid=24557.

Rösch, Daniel, and Harald Scheule. 2008. Stress Testing for Finan-cial Institutions: Applications, Regulations, and Techniques. Lon-don: Risk Books, Incisive Media.

Schmieder, Christian, Claus Puhr, and Maher Hasan. 2011. “Next Generation Balance Sheet Stress Testing.” IMF Working Paper 11/83, International Monetary Fund, Washington, DC. https://w w w.imf.org/en/Publ icat ions/W P/Issues/2016/12/31 /Next-Generation-Balance-Sheet-Stress-Testing-24798.


https://www.imf.org/external/pubs/cat/longres.aspx?sk=25509.0


https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Banking-Stability-Measures-22554



http://eprints.lse.ac.uk/24511/

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Financial-Contagion-Through-Bank-Deleveraging-Stylized-Facts-and-Simulations-Applied-to-the-24285




http://bruegel.org/2011/07/stress-tests-fail-to-rescue-europes-banks/

http://bruegel.org/2011/07/stress-tests-fail-to-rescue-europes-banks/

https://www.hkma.gov.hk/eng/publications-and-research/research/working-papers/2009/./en/Publications/WP/Issues/2016/12/31/Revisiting-Risk-Weighted-Assets-25807




https://www.financialresearch.gov/annual-reports/2012-annual-report/


https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Of-Runes-and-Sagas-Perspectives-on-Liquidity-Stress-Testing-Using-an-Iceland-Example-24019



http://www.mdpi.com/2227-7072/2/1/15

https://rbi.org.in/Scripts/BS_PressReleaseDisplay.aspx?prid=24557

https://rbi.org.in/Scripts/BS_PressReleaseDisplay.aspx?prid=24557

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Next-Generation-Balance-Sheet-Stress-Testing-24798



CHAPTER 3

Stress Tests as a Systemic Risk Assessment Tool

DIMITRI G. DEMEKAS

Turning stress tests into a useful tool for assessing system- wide risk requires (1) incorporating general equilibrium dimensions, so that the out-come of the test depends not only on the size of the shock and the initial buffers of individual institutions but also on their responses to the

shock and their interactions with each other and with other economic agents; and (2) focusing on the resilience of the system as a whole. Progress has been made toward the first goal: several models are now available that capture behavioral responses and feedback effects. But building mod-els that measure correctly systemic risk and the contribution of individual institutions to it has proved more difficult. Further progress in this area would entail using a variety of analytical approaches and scenarios, integrating non bank financial entities, and exploring the use of agent- based models. As well, stress tests should not be used in isolation, but be treated as complements to other tools and— crucially— be combined with microprudential perspectives.

Federal Reserve Chairman Ben Bernanke, regulators and policymakers need a “broader field of vision” (Bernanke 2008). One of the early lessons from the crisis was thus the need to adapt and broaden the “traditional” risk- monitoring toolkit to capture systemic risk better.

Stress testing has long been one of the key tools for assess-ing risks and resilience in financial institutions. The global financial crisis sparked a renewed interest in stress tests, which have now become a prominent— and in some cases statutory— feature of the regulatory regimes in many juris-dictions. Stress testing, once an arcane subject, has become a household term. But how well suited are stress tests in assess-ing systemic risk in the financial sector? How far has this “broader field of vision” been adopted in the current stress testing models? And where do we go from here?

These are the central questions this chapter is trying to address. Section 2 provides a brief primer on stress testing and how it has evolved over time. Section 3 discusses the challenge of adapting stress testing models to the task of as-sessing systemic risk and provides a critical evaluation of the

1. INTRODUCTIONThe importance of systemic risk originating in a country’s financial sector— in other words, the risk of a shock that dis-rupts the functioning of the financial sector to such an ex-tent as to have major consequences on the rest of the economy— was dramatically highlighted during the global financial crisis. To be sure, the notion that a stock market crash or a banking crisis can have knock- on effects on the real economy is not new: economic history has plenty of ex-amples of both. But the global financial crisis underscored three new elements. First, the degree of systemic risk is not just a function of the magnitude of vulnerabilities facing in-dividual sectors but also of the interconnections between them, as well as between financial firms, nonfinancial firms, and consumers, both domestically and across borders. Sec-ond, even the failure of individual firms can create systemic risk if these are too big or too interconnected to fail. And third, traditional policy frameworks that rely only on mon-etary, fiscal, and microprudential policies are not sufficient to contain systemic risk. In the words of the previous US

This chapter appeared originally as Demekas 2017, and is reproduced here by permission. This chapter is based largely on the findings of a longer piece of research by the author (Demekas 2015). Comments by a number of colleagues at the IMF, especially Jorge Chan- Lau, Laura Valderrama, and Miguel Segoviano, as well as participants to the 6th Expert Forum on Advanced Stress Testing Techniques at the Bank of England’s Centre for Central Banking Studies in December 2014 and in the Prudential Regulation Authority’s International Regulatory Risk Roundtable in April 2015, are gratefully acknowl-edged, but all remaining errors are the author’s responsibility. The views expressed herein are those of the author and should not be attributed to the IMF, its Executive Board, or its management.


56 Stress Tests as a Systemic Risk Assessment Tool

focused on safeguarding the safety and soundness of indi-vidual institutions. And this, it was thought, would ensure the stability of the financial system as a whole.

3. A NEW GENERATION OF STRESS TESTSEven as bank regulators were putting the finishing touches on the Basel II framework, many understood that ensuring the safety and soundness of each individual institution was neither necessary nor sufficient to ensure that the financial system as a whole would remain stable and continue to func-tion. As Andrew Crockett, then General Manager of the Bank for International Settlements, put it: the micropruden-tial approach to financial regulation may “strive for too much and deliver too little” (Crockett 2000). It may strive for too much because the occasional failure of individual in-stitutions is not the problem, if other institutions are capable of stepping in and providing intermediation services. Trying to avoid such outcomes risks providing “excessive protec-tion.” And it may deliver too little because it does not take into account how each individual institution pursues com-pliance with capital regulation. When, for example, a regu-lator pushes a troubled bank to restore its capital ratio, the regulator does not care whether the bank increases capital or shrinks assets. But if a substantial proportion of the banking system shrinks assets simultaneously to meet capital require-ments, the damage to the economy may be considerable. Unless regulators take into account the collective behavior of institutions in response to a shock (or to regulatory require-ments), they may fail to minimize the probability of distress for the system as a whole and the associated economic costs— in short, systemic risk.

Moving from the traditional microprudential stress tests toward a new generation of stress tests that, in Andrew Crockett’s words, would “marry the microprudential and macroprudential dimensions of financial stability” involves two challenges:

• Introducing general equilibrium dimensions. This does not require that the economy or the financial system are at equilibrium at any given time, it just means that the stress tests are designed so that the outcome depends not only on the size and nature of the initial shock and the buffers of individual finan-cial institutions, but also on the behavioral responses of these institutions as the shock unfolds, and on the interactions of these institutions with each other and with other economic agents (borrowers, funding providers, depositors, and economic policymakers).

• Shifting the focus of the stress tests from individual institutions to the resilience of the system as a whole, in other words, on its ability to continue functioning and providing financial intermediation services to the economy.

How much progress have stress testers made in tackling these challenges?

progress made so far. And Section 4 outlines the key priori-ties in order to complete this task.

One important caveat is in order. This chapter is not a comprehensive overview of stress testing practices around the world. It does not provide advice on how to run “better” stress tests overall. The focus is solely on the use of stress tests for systemic risk assessment. Several important aspects of stress testing, like scenario selection and design, data quality, shock calibration, communication of results, or policy follow-up, are not covered. Fortunately, there is no shortage of studies covering these other aspects of stress tests.

2. “PLAUSIBLE, SEVERE, AND RELEVANT”: THE ORIGINS OF STRESS TESTS FOR BANKSStress testing is not a recent invention. Originally used in engineering, stress analysis is a technique for testing a struc-ture or system beyond normal operating capacity, often to the breaking point, to confirm specifications are met, deter-mine breaking limits, or examine modes of failure. Asset managers and financial institutions, as well as their supervi-sors, have also realized the benefit of submitting portfolios or entire balance sheets to numerical simulations of hypo-thetical shocks to selected variables, like various asset prices, and assessing the impact on profits, capital, or the ability of regulated institutions to continue meeting their obligations, including observing regulatory requirements.

One of the early adopters of stress tests in the early 1990s was J.P. Morgan, whose RiskMetrics methodology used value- at- risk to measure market risk: in other words, the po-tential loss over a specific time period from movements in asset prices with a certain probability (see Zangari 1996). Regulators caught up after a while, and the Basel II capital framework required banks to perform stress tests for market risk and, in some cases, credit risk. These tests had to be “plausible, severe, and relevant” to help the bank evaluate its capacity to absorb losses and identify steps it can take to re-duce risk and conserve capital (BCBS 2005).

Early stress testing models were relatively simple. They assumed a hypothetical shock— for instance, on credit qual-ity or asset prices— and calculated the associated losses, making simplistic assumptions about the behavior of the bank (on profit distribution, credit expansion or deleverag-ing, and so on). They focused on the solvency of the indi-vidual bank, that is, on the impact of the shock on the bank’s capital. Liquidity risk was treated separately from solvency, if at all, and interactions among banks were generally ignored.

These stress tests had a microprudential focus. Their ob-jective was to assess the capacity of an individual institution to absorb losses under adverse conditions, that is, its ability to avoid failure and continue to function and to meet regulatory requirements. This was consistent with the domi-nant approach to financial regulation at the time, which was


Dimitri G. Demekas 57

can in theory capture all sources of vulnerability and contagion, including the risk of self- fulfilling runs triggered by investor sentiment that may not neces-sarily reflect weak fundamentals. Another advantage of these models is computational simplicity: com-bined with the availability of very high frequency market data, this makes them ideal for high fre-quency monitoring of bank resilience to a variable set of risks. – An obvious weakness of these models is their reli-

ance on market data, which are noisy and may overestimate or underestimate risks. Bank default risk indicators estimated from these data may thus be excessively volatile and may not provide a sound basis for bank management or supervisory action. Another pitfall is that by extracting infor-mation from market data and constructing a summary metric of bank soundness, market- price- based models do not allow the stress tester to differentiate between the various factors that contribute to the bottom line (initial shock, risk inter dependence, common exposures, cross- institution contagion, or market sentiment). In-stead, all these factors are lumped into the implied probability of default or distress generated by the model. This had led some critics to dismiss these models as “black boxes.”

In contrast to the progress made toward incorporating general equilibrium dimensions into the traditional micro-prudential stress testing framework, relatively little has been done to tackle the second challenge. Few stress testing mod-els focus on— and measure correctly— the resilience of the financial system as a whole and its ability to continue pro-viding financial intermediation services under stress. This reflects two fundamental problems: the aggregation problem and the robustness problem.

• The aggregation problem reflects the fact that the sum of individual banks’ losses or capital shortfalls in the event of a shock is not a good proxy of system- wide risk. Given the different ways in which banks are in-terconnected, the individual losses are not additive. Correctly aggregating individual losses requires some knowledge of the dependence structure between indi-vidual bank balance sheets. And this dependence structure is nonlinear and tends to vary with the de-gree of systemic stress—that is, to increase at times of stress.

• The robustness problem expresses the notion that a single stress scenario, however severe, does not pro-vide enough information about the resilience of the system to other shocks with the same probability. A theoretically more correct approach would be to look simultaneously at all risk factors affecting the sys-tem; estimate a multidimensional region with a given probability mass— say 95 or 99 percent; and calculate the maximum loss of the system for all

On the basis of a review of the experience of central banks, supervisory institutions, macroprudential authori-ties, and the IMF with stress testing models (Demekas 2015), the answer is: quite a lot, but we are not there yet. The stress testing community has made significant progress in tackling the first of the two challenges but much less in deal-ing with the second.

A number of models that incorporate some general equi-librium dimensions into stress tests are now available and widely used. They fall into two broad categories, each with its own strengths and pitfalls.

• Balance-sheet-based models that, as the name im-plies, use individual bank balance sheet data to as-sess the impact of an exogenous shock on asset quality, income, and ultimately capital (for sol-vency tests) or various measures of liquidity (for li-quidity tests). The results are then aggregated to give an idea of the vulnerability of the system as a whole. In this approach— by far the most common across central banks and supervisory agencies around the world— any general equilibrium di-mensions the stress tester intends to capture, whether solvency- liquidity interactions, behavioral responses, or macro feedback effects, are built in the model. – This benefit, however, comes at a price. First, these

models can only capture general equilibrium ef-fects that are explicitly incorporated in the frame-work. Second, analytical and computational complexity and data requirements increase very rapidly as more features are added to the models. This renders them slow, cumbersome, and costly to construct and run. Third, since they rely on bank balance sheet data, they depend crucially on the availability and quality of these data. Fourth, the microfoundations of these models are weak: instead of an integrated framework of optimizing economic agents, most of these models use ad hoc mechanisms and rules- of- thumb to incorporate dynamic effects and behavioral responses. Argu-ably, this is a price worth paying for introducing, in a practical way, general equilibrium dimensions in the traditional balance-sheet-based stress test-ing models. Moreover, one could point out that these behavioral rules of thumb are not arbitrary but derived from observations of past behavior. Nevertheless, this is an important handicap of these models, especially in times of crisis, when observed behavioral patterns break down and economic agents learn and adapt, often on the ba-sis of imperfect information. And lastly, these models cannot capture the propagation of risks through market contagion, credibility effects, and so on.

• Models using (mostly) market data to infer the prob-ability of distress or default of individual institutions



4. THE WAY FORWARDHow do we move toward making stress tests more effective tools for systemic risk assessment? What follows is not a sys-tematic research agenda, but a set of practical suggestions for stress testing practitioners that, in the author’s view, can yield improvements or address obvious pitfalls and, in this way, move the dial on system- wide stress tests.

Use a Variety of Models

Given the limitations of the existing stress testing frame-works, it is surprising to see several central banks and regula-tory agencies relying on a single model. This makes the outcome of the stress test hostage to the limitations of a single analytical framework.

Instead, a variety of models should be used for system- wide stress testing. The challenge in this case would be to in-terpret and synthesize the results of the different models into a coherent and persuasive narrative. Should the different re-sults be combined or averaged according to a strict rule? Should qualitative judgment be used in weighing different— and potentially contradictory— results? These are not easy questions. But this is a challenge well worth tackling, as it would enhance the insights into systemic risk and the quality of the ensuing conversation about financial stability, both within the supervisory agency and with the banks.

Run More— and Smarter— Stress Scenarios

Most stress testing exercises are limited to one or two macro-economic stress scenarios (for instance, an “adverse” and a “severe” recession scenario). This approach has a major pit-fall: resilience to a shock of a given probability does not im-ply resilience to all shocks with the same probability (the robustness problem). It also ignores the increasingly impor-tant cross- border nature of risk: banks and other financial institutions are increasingly interlinked across borders, and may be vulnerable to shocks that originate in— or propagate through— a foreign country or market. The outcome of a single stress scenario focused on a domestic recession may thus be misleading.

The obvious remedy is to use a multitude of extreme but plausible scenarios for the stress tests. This would provide a better sense of the resilience of the system and of individual institutions to a range of shocks than a single scenario. Us-ing multiple scenarios (as well as a variety of models) would also have another big advantage: it would minimize the scope for individual institutions to “game the test,” that is, make portfolio choices geared toward passing the specific stress test— a risk that was recognized early on (Office of Financial Research 2012; Bank of England 2013).

In addition to the number, a related issue is the type of scenarios used in stress tests. In most cases, the main stress scenario is an adverse macroeconomic shock exogenous to

scenarios falling in this region.1 This would measure the resilience of the system to all plausible scenarios with a probability of at least 95 or 99 percent. But this approach is very hard to implement and has not so far been used in stress tests conducted by a major central bank.

Both problems have long been recognized but are tough to crack. Some market-price- based models do not face the aggregation problem: models like Adrian and Brunnermei-er’s CoVar (Adrian and Brunnermeier 2016) or Segoviano and Goodhart’s distress dependence (Segoviano and Good-hart 2010) start by estimating systemic risk and then derive the individual bank’s contribution to it. One issue, however, with these models is that they do not translate these risk metrics into something that can be readily compared to the regulatory capital or liquidity requirements. Since these are ultimately the main tools microprudential or macropruden-tial regulators can use to mitigate risk, both at the individual bank and at the systemic level, a model without this element is unlikely to be of much practical value to policymakers. As regards the robustness problem, despite some promising the-oretical approaches,2 no satisfactory practical solution has been widely accepted yet.

1 Estimating the maximum loss for a multidimensional region of a given probability mass has an undesirable property known as the dimensional dependence of maximum loss. As an example, start with a bond portfo-lio with risk factors consisting of two yield curves in 10 currencies. One risk manager models the yield curve using seven maturity buckets and another using 15 maturity buckets. Both choose a plausibility region of 95 percent. It has been shown (Breuer 2008) that the second risk man-ager will calculate a maximum loss that is 1.4 times higher than that calculated by the first risk manager, although both look at the same portfolio and the same plausibility level. Addressing this technical pit-fall requires using a slightly different statistical concept of plausibility (Breuer and others 2009).

2 One such approach is proposed in Webber and Willison (2011). Instead of estimating system- wide losses under a stress scenario, distributing those losses among individual banks, and then comparing the outcome to starting bank capital, this approach recasts the problem entirely from the policymaker’s perspective. The policymaker is interested in setting individual bank capital at a level that ensures systemic solvency over a given time horizon at a certain level of probability. Given the trade- off between stability and efficiency, this is set up as a constrained optimiza-tion problem, where bank capital requirements are minimized subject to a specified probabilistic systemic stability target. Another promising theoretical approach that tackles both the aggregation and the robust-ness problem is outlined in Pritsker 2014. The key innovation here is a definition of systemic stability that is directly related to the level of regulatory capital of each bank: the “system assets in distress” (SAD) measure is defined as the sum of each bank’s intermediation capacity, in turn defined as a function of its total assets and capital times a vector of risk factors. A continuous measure of systemic risk is then given as the probability that SAD exceeds a prespecified level θ (p(SAD > θ)) for a given time horizon. The model then sets up a constrained stress maxi-mization problem to estimate the amount of capital needed for each bank so as to satisfy the constraint that p(SAD > θ) ≤ α, where θ is the regulator’s systemic risk target. The constrained stress maximization uses Monte Carlo simulations to estimate nonparametrically the prob-ability density function of SAD so as to take into account distress de-pendence and cover all possible realizations of the risk factors at a certain probability level.


Dimitri G. Demekas 59

• The emergence of new dynamics under stress, when relationships among financial institutions can change suddenly3

• The fact that the response of regulated institutions to policy signals is state- contingent (for example, rais-ing the regulatory capital requirements put in place in normal times to ensure banks have sufficient capi-tal buffers may have no positive effect on systemic stability at times of crisis [Klinger and Teplý 2014])

Agent- based models can capture many of these aspects, and may be better suited for analyzing situations of financial stress.4 An agent- based model postulates autonomous, het-erogeneous agents with bounded rationality, and specifies heuristic rules that dictate how they will act based on various factors. These rules can vary across different types of agents (for instance banks, depositors, providers of wholesale fund-ing) and allow for less- than- optimal behavior. The model de-termines the “topology,” that is, the mechanism through which agents can interact (for example, how they form net-works), and can explore various types of shocks, both exoge-nous and endogenous (such as changes in agent behavior, topology rules, and so on). Agent- based models are increas-ingly being used for macro- financial modeling, and relatively simpler versions have been used to explore the impact of stress scenarios on bank solvency, liquidity, and contagion.5 Agent- based models are complex, and implementing them would require a shift in the approaches traditionally taken by (and the skills traditionally required of) stress testers. Neverthe-less, the limited experience so far suggests that they can pro-vide unique insights into the aspects that matter most in a crisis: the behavioral responses of banks and the interactions between banks, market participants, and policymakers.

Embed Stress Tests into the Financial Stability Policy Framework

The recent explosion of interest in stress testing is creating a risk. Policymakers, market participants, and the broader pub-lic may focus excessive attention on stress tests, form exagger-ated expectations, take stress test results out of context, and give them much greater weight than they merit in guiding policy action. This risk is evident in the way stress test results tend to dominate the public debate on the health of the banks in the United States following adoption of the Dodd- Frank Act, as well as in Europe following a string of highly publicized

3 For example, the endogenous network literature has explored how net-work formation changes depending on the environment— see Deb 2012.

4 This point was argued compellingly by Bookstaber (2012).5 A well- known example is the Complexity Research Initiative for Sys-

temic Instabilities, building a large- scale macro- financial agent- based model for the European economy. A relatively simpler version is used by Chan- Lau (2014) to explore the impact of financial regulation on bank solvency, liquidity, and contagion under stress scenarios. This paper also contains a brief literature survey on the use of agent- based models for macro- financial modeling.

the financial sector. But in many actual financial crises, the shock originates entirely inside the financial system and is then followed by a recession. A well- known study of 43 banking crises in 30 countries (Alfaro and Drehmann 2009) shows that only about half were preceded by adverse macro-economic conditions.

Effective system- wide stress tests should therefore involve a higher number and a wider range of “smart” stress scenar-ios. This is not a new idea, but implementation is demanding. It requires an in- depth understanding of the risks affecting the financial system, including cross- border dimensions. It would also complicate the task of synthesizing and commu-nicating the results— especially when accompanied by a vari-ety of models—and would cost more. It is these challenges that have held back many supervisors from moving in this direction. However, given the significant pitfalls of limiting the number of scenarios to just one or two, it may be time to reconsider the cost- benefit balance of the current approach.

Expand Coverage to Nonbank Financial Entities

Microprudential stress tests have been traditionally applied to banks because these were the predominant agents of financial intermediation. Recent trends have undermined this fact. The line between banks and nonbanks has been blurred. The nonbanking industry has expanded greatly in size and im-portance in the last two decades and the global financial cri-sis has demonstrated that banks and nonbanks are deeply interconnected and risks move easily between the two. There-fore, from a systemic perspective, stress tests should cover both banks and nonbanks, with the choice of which nonbank entities to incorporate into the stress testing framework de-pending on country circumstances. Priority should be given to sectors that are closely connected with banks through ownership or financial linkages— typically insurance compa-nies. Asset management companies, mutual funds, and sometimes pension funds are also sometimes important pro-viders of liquidity to banks, and could be affected by— or be a propagation channel for— a systemic shock.

Explore Agent- Based Models

Stress testing models— like all traditional economic and fi-nancial models based on neoclassical microfoundations— face a fundamental problem: they assume homogeneous agents (individuals or institutions) that always behave ratio-nally in ways that can be modeled based on past experience, and that policy decisions influence this behavior in the same way for all market participants. These assumptions miss some critical points about financial crises, notably:

• The fact that that market participants, both in the financial and the nonfinancial sectors, are heteroge-neous and often make less- than- rational decisions, especially under stress



Federal Reserve. https://www.federalreserve.gov/newsevents /speech/bernanke20080822a.htm.

Bookstaber, Richard. 2012. “Using Agent-Based Models for Ana-lyzing Threats to Financial Stability.” Office of Financial Re-search Working Paper 0003, US Department of the Treasury, Washington, DC. https://www.financialresearch.gov/working papers/2012/12/21/using-agent-based-models-for-analyzing -threats-to-financial-security/.

Breuer, Thomas. 2008. “Overcoming Dimensional Dependence of Worst Case Scenarios and Maximum Loss.” Journal of Risk 11 (1): 79–92.

Breuer, Thomas, Martin Jandacka, Klaus Rheinberger, and Martin Summer. 2009. “How to Find Plausible, Severe, and Useful Stress Scenarios.” International Journal of Central Banking 5 (3): 205–224.

Chan-Lau, Jorge A. 2014. “Regulatory Requirements and Their Implications for Bank Solvency, Liquidity, and Interconnected-ness Risks: Insights from Agent-Based Model Simulations.” Social Sciences Research Network, https://papers.ssrn.com /sol3/papers.cfm?abstract_id=2537124.

Crockett, Andrew D. 2000. “Marrying the Micro- and Macro-Prudential Dimensions of Financial Stability.” Remarks before the 11th International Conference of Banking Supervisors, Ba-sel, September 20–21, Bank of International Settlements. http://www.bis.org/speeches/sp000921.htm.

Deb, Pragyan. 2012. “Market Frictions, Interbank Linkages, and Excessive Interconnections.” Social Sciences Research Network. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2121923.

Demekas, Dimitri G. 2015. “Designing Effective Macroprudential Stress Tests: Progress So Far and the Way Forward.” IMF Working Paper 15/146, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Is sues /2016/12/31/Designing-Effective-Macroprudential-Stress-Tests -Progress-So-Far-and-the-Way-Forward-43043.

———. 2017. “Stress Tests as a Systemic Risk Assessment Tool.” Journal of Risk Management in Financial Institutions 10 (1): 36–44.

International Monetary Fund (IMF). 2012. “Macro-financial Stress Testing: Principles and Practices.” IMF Policy Paper, Washington, DC. https://www.imf.org/en/Publications/Policy -Papers/Issues/2016/12/31/Macrofinancial-Stress-Testing -Principles-and-Practices-PP4702.

Klinger, Tomas, and Petr Teplý. 2014. “Systemic Risk of the Global Banking System—An Agent-Based Network Model Approach.” Prague Economic Papers 1: 24–41.

Office of Financial Research. 2012. 2012 Annual Report. US De-partment of the Treasury, Washington, DC. https://www .financialresearch.gov/annual-reports/2012-annual-report/.

Pritsker, Matt. 2014. “Enhanced Stress Testing and Financial Sta-bility.” Social Sciences Research Network. https://papers.ssrn .com/sol3/papers.cfm?abstract_id=2082994.

Segoviano, Miguel A., and Charles E. Goodhart. 2010. “Distress Dependence and Financial Stability.” In Financial Stability, Monetary Policy, and Central Banking, edited by Rodrigo Alfaro. Santiago: Central Bank of Chile.

Webber, Lewis, and Matthew Willison. 2011. “Systemic Capital Requirements.” Bank of England Working Paper 436, London. https://www.bankofengland.co.uk/working-paper/2011 /systemic-capital-requirements.

Zangari, P. 1996. “Statistical and Probability Foundations.” In RiskMetricTM—Technical Document, 4th edition. New York: Morgan Guarantee Trust Company of New York.

tests originally by the European Banking Authority and now the European Central Bank. Immediate remedial action by bank management and supervisors is automatically expected— or indeed required— of banks “failing” the test.

This was less of an issue with the traditional micropru-dential stress tests. Those tests were an input into the assess-ment of the soundness of individual institutions. Their results were not made public and rarely triggered automatic remedial action; they were instead used to inform the ongo-ing conversation between the regulated entity and the regu-lator. But the unprecedented attention focused on recent system- wide stress tests in advanced economies seems at times to overshadow, rather than inform, the conversation about financial stability among policymakers, regulators, in-dividual institutions, market participants, and the public.

This risk has been noted before. In setting out best- practice principles for stress testing, the IMF put it this way (IMF 2012):

Regardless of how extensive the coverage of risk factors, how refined the analytical models, how severe the shocks incorporated in the stress tests, and how careful the com-munications strategy, there is always the risk that the ‘un-thinkable’ will materialize. […] No matter how hard the stress tester tries, stress tests always have margins of error. Their results will almost always turn out to be optimistic or pessimistic ex post. In addition, there will always be model risk, imperfect data, or underestimation of the se-verity of the shock. One should therefore set stress test results in a broader context.

So the call to embed stress tests firmly in the financial stability framework is essentially a call for caution and hu-mility. Stress testing is just one of the many tools available to assess systemic vulnerabilities and resilience. They should be treated as complements to other tools, such as early warning indicators, and— crucially— should be combined with mi-croprudential perspectives.

REFERENCESAdrian, Tobias, and Markus Brunnermeier. 2016. “CoVaR.” Amer-

ican Economic Review 106 (7): 1705–41.Alfaro, Rodrigo, and Mathias Drehmann. 2009. “Macro Stress

Tests and Crises: What Can We Learn?” Bank of International Settlements Quarterly Review (December). https://www.bis.org /publ/qtrpdf/r_qt0912e.htm.

Bank of England. 2013. “A Framework for Stress Testing the UK Bank-ing System.” Bank of England Discussion Paper, October, London. https://www.bankofengland.co.uk/paper/2013/a-framework -for-stress-testing-the-uk-banking-system.

Basel Committee on Banking Supervision (BCBS). 2005. “Inter-national Convergence of Capital Measurement and Capital Standards: A Revised Framework.” Technical Report, Bank for International Settlements, Basel, Switzerland. https://www.bis .org/publ/bcbs128.htm.

Bernanke, Ben S. 2008. “Reducing Systemic Risk.” Remarks at the Federal Reserve Bank of Kansas City’s Annual Economic Sym-posium, Jackson Hole, Wyoming, Board of Governors of the


https://www.federalreserve.gov/newsevents/speech/bernanke20080822a.htm


https://www.financialresearch.gov/working

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2537124


http://www.bis.org/speeches/sp000921.htm

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2121923

https://www.imf.org/en/Publications/WP/Is

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Macrofinancial-Stress-Testing-Principles-and-Practices-PP4702







https://www.bankofengland.co.uk/working-paper/2011/systemic-capital-requirements

https://www.bankofengland.co.uk/working-paper/2011/systemic-capital-requirements

https://www.bis.org/publ/qtrpdf/r_qt0912e.htm


https://www.bankofengland.co.uk/paper/2013/a-framework-for-stress-testing-the-uk-banking-system

https://www.bankofengland.co.uk/paper/2013/a-framework-for-stress-testing-the-uk-banking-system





PART II

Concepts


CHAPTER 4

The Global Macro- Financial Model: A Stress Test Scenario Simulation Tool

FRANCIS VITEK

This chapter presents the global macro- financial model (GFM) and discusses its application to simulating banking sector solvency stress test scenarios. The GFM is a structural macroeconometric model of the world economy, disaggregated into 40 national economies, which features a

range of nominal and real rigidities, extensive macro- financial linkages, and diverse spillover transmission channels. The simulation of a stress test scenario using the GFM is demonstrated with an example.

model of the world economy, disaggregated into 40 national economies within a panel framework. This estimated New Keynesian DSGE model features a range of nominal and real rigidities, extensive macro- financial linkages, and di-verse spillover transmission channels. These macro- financial linkages encompass bank- and capital-market- based finan-cial intermediation, with cross- border balance sheet expo-sures and contagion effects. These features enable the GFM to simulate diverse stress test scenarios, generating theoreti-cally coherent and empirically adequate dynamic interrela-tionships across macro- financial variables driven by sets of shocks having structural interpretations, potentially subject to policy constraints. These scenarios internalize key macro- financial feedback loops, in particular between the banking sector and the rest of the economy, while capturing its direct and indirect exposures to risks originating domestically or abroad.

The organization of this chapter is as follows. Section 2 presents the equations governing the evolution of the endog-enous and exogenous variables of the GFM. Estimation of the model is the subject of Section 3. The simulation of stress test scenarios using the GFM is discussed in Section 4, with reference to an example for the United Kingdom. Finally, Section 5 offers conclusions and outlines future model devel-opment plans.

2. THE MODELThe theoretical foundation of the GFM is the canonical New Keynesian DSGE model of an open economy, extended to incorporate additional macro- financial linkages and spillover transmission channels, while preserving analytical and com-putational tractability. Following Smets and Wouters 2003,

1. INTRODUCTIONBanking sector solvency stress tests are generally based on severe but plausible macro- financial scenarios. These scenar-ios specify the future evolution of a variety of macroeco-nomic and financial market variables, given sets of shocks that perturb them from their baseline paths, as well as feasi-ble policy responses. These sets of shocks represent scenario narratives capturing key macro- financial risks to which the banking sector is exposed, and typically have structural in-terpretations. These risks may originate domestically or abroad, with the banking sector exposed to them either di-rectly via determinants of profitability and capitalization such as funding costs and loan impairments, or indirectly through macro- financial linkages to these determinants.

Estimated dynamic stochastic general equilibrium (DSGE) models are widely used by monetary and fiscal authorities for policy analysis and forecasting purposes because they can generate theoretically coherent and empirically adequate dy-namic interrelationships across macro- financial variables, driven by shocks having structural interpretations. This class of structural macroeconometric models has many variants, with the estimated New Keynesian DSGE models used by policymaking institutions incorporating a range of nominal and real rigidities, as well as an expanding array of macro- financial linkages. These properties make New Keynesian DSGE models well suited for simulating stress test scenarios, in particular those that feature a banking sector, as this inter-nalizes key macro- financial feedback loops.

This chapter presents the global macro- financial model (GFM), documented more fully in Vitek 2015, and discusses its application to simulating banking sector solvency stress test scenarios. The GFM is a structural macroeconometric


The Global Macro-Financial Model: A Stress Test Scenario Simulation Tool64

rational expectation of variable xi,t+s conditional on informa-tion available in period t. Bilateral weights ,wi j

Z for evaluating the trade weighted average of variable xi,t across the trading partners of economy i are based on exports for Z = X, imports for Z = M, and their average for Z = T. In addition, bilateral weights ,wi j

Z for evaluating the weighted average of variable xi,t across the lending destinations and borrowing sources of economy i are based on bank lending for Z = C and nonfinan-cial corporate borrowing for Z = F. Furthermore, bilateral weights ,wi j

Z for evaluating the portfolio weighted average of variable xi,t across the investment destinations of economy i are based on debt for Z = B and equity for Z = S. Finally, world weights wi

Z for evaluating the weighted average of variable xi,t across all economies are based on output for Z = Y, money market capitalization for Z = M, bond market capital-ization for Z = B, and stock market capitalization for Z = S.

Endogenous Variables

Output price inflation ˆ ,i tYπ depends on a linear combination

of its past and expected future values driven by the contem-poraneous labor income share, output, and internal terms of trade according to the output price Phillips curve:

∑

π γγ β

π βγ β

π ω ω βω γ β

φθ

θ

=+

++

+− −

+

+ − −

+ −−

− +

=

−

ˆ1

ˆ1

E ˆ(1 )(1 )

(1 )ln

ˆ ˆ

ˆ ˆ

1 1 ln ˆ ln ˆ 1

1ln ˆ

, , 1 , 1, ,

, ,

,

1

*1

, , ,

W L

P Y

X

Y

X

XY

X

Y

i tY

Y

Y i tY

Y t i tY

Y Y

Y Y

i t i t

i tY

i t

i

i

i k

ikF

k

M

i ti

ii t

XY i t

Y

P T+ ∆( ) ln ˆ .1 ,

X

YLi

ii t

X (4.1)

Output price inflation also depends on contemporaneous, past and expected future changes in the internal terms

of trade, where the polynomial in the lag operator

( ) 111 = − γ

γ β+L L

Y

Y − Εβγ β+

−

11LY t . The response coefficients

of this relationship vary across economies with their trade

openness and commodity export intensities.Consumption price inflation ˆ ,i t

Cπ depends on a linear combination of its past and expected future values driven by the contemporaneous labor income share, output, and the internal terms of trade according to the consumption price Phillips curve:

∑

π γγ β

π βγ β

π ω ω βω γ β

φθ

θ

=+

++

+− −

+

+ − −

+ −−

− +

=

−

ˆ1

ˆ1

ˆ(1 )(1 )

(1 )ln

ˆ ˆ

ˆ ˆ

1 1 ln ˆ ln ˆ 1

1ln ˆ

E, , 1 , 1, ,

, ,

,

1

*1

, , ,

W L

P Y

X

Y

X

XY

X

Y

i tC

Y

Y i tC

Y t i tC

Y Y

Y Y

i t i t

i tY

i t

i

i

i k

ikF

k

M

i ti

ii t

XY i t

Y

P T+ ∆( ) ln ˆ .1 ,

M

YLi

ii t

M (4.2)

Consumption price inflation also depends on contempora-neous, past, and expected future changes in the external terms of trade. The response coefficients of this relationship

the GFM features short- term nominal price and wage rigidi-ties generated by monopolistic competition, staggered reopti-mization, and partial indexation in the output and labor markets. Following Christiano, Eichenbaum, and Evans 2005, the resultant inertia in inflation and persistence in out-put is enhanced with other features such as habit persistence in consumption, adjustment costs in investment, and variable capital utilization. Following Galí 2011, the model incorpo-rates involuntary unemployment through a reinterpretation of the labor market. Households are differentiated according to whether they are bank intermediated, capital market inter-mediated, or credit constrained. Bank-intermediated house-holds have access to domestic banks where they accumulate deposits, whereas capital- market- intermediated households have access to domestic and foreign capital markets where they trade financial assets. Motivated by Tobin 1969, these capital- market- intermediated households solve a portfolio balance problem, allocating their financial wealth across do-mestic and foreign money, bond and stock market securities that are imperfect substitutes. To cope with the curse of di-mensionality, targeted parameter restrictions are imposed on the optimality conditions determining the solution to this portfolio balance problem, avoiding the need to track the evolution of granular asset allocations. Firms are grouped into differentiated industries. The energy and nonenergy commodity industries produce internationally homogeneous goods under decreasing returns to scale, while all other in-dustries produce internationally heterogeneous goods under constant returns to scale. Banks perform global financial in-termediation subject to financial frictions and a regulatory constraint. Building on Hülsewig, Mayer, and Wollmershäuser 2009, they issue risky domestic currency- denominated loans to domestic and foreign firms at infrequently adjusted prede-termined lending rates. Also building on Gerali, Neri, Sessa, and Signoretti 2010, they obtain funding from domestic bank- intermediated households via deposits and from the domestic money market via loans, accumulating bank capi-tal out of retained earnings given credit losses to satisfy a regulatory capital requirement. Motivated by Kiyotaki and Moore 1997, the GFM incorporates a financial accelerator mechanism linked to collateralized borrowing. Finally, fol-lowing Monacelli 2005, the model accounts for short- term incomplete exchange rate pass-through with short- term nom-inal price rigidities generated by monopolistic competition, staggered reoptimization, and partial indexation in the im-port markets.

Simulation is based on an estimated linear state space rep-resentation of an approximate multivariate linear rational ex-pectations representation of this New Keynesian DSGE model of the world economy. This multivariate linear rational expectations representation is derived by analytically lineariz-ing the equilibrium conditions of the model around its sta-tionary deterministic steady state equilibrium, and consolidating them by substituting out intermediate variables. In what follows, ˆ ,xi t denotes the deviation of variable xi,t from its steady state equilibrium value xi, while Et xi,t+s denotes the


Francis Vitek 65

vary across economies with their trade openness and com-modity export intensities.

Output ln ˆ,Yi t depends on a weighted average of its past

and expected future values driven by a weighted average of the contemporaneous real ex ante portfolio return and short- term real market interest rate according to output demand relationship:

P P

P T T∑

αα α

φ σ αα

φφ

φφ

νν

φ Π Πτ

ττ

νν ν

ψ

=+

++

− −

−−+ −

+ −−

−

− + − −−

− +

+ − −

− −

− + ++

=

ln ˆ1

ln ˆ 1

1E ln ˆ 1 (1 )

1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ

( ) lnˆ ˆ

ˆ ˆ1 ln ˆ 1 ln ˆ .

, , 1 , 1 , 1 ,,

, 1

2,

,

, ,

,, 2 , ,

2 ,, ,

, ,, ,

1

,Y Y YX

Y

C

Yr r

LP Y P

W L

P Y

W L

PL

I

YI

G

YG

X

YL w

D M

Y

M

Y

i t i t t i ti

i

C i

it

A

C i tA

A

C i tS i t

C

i tC

C iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

i

ii jX i t

Mj t

i tX

j tM

M j

jj tM i

ii t

M

j

N

A H

T

φ σ αα

φφ

φφ

νν

ττ

τ

νν ν

ψ

−−+ −

+ −−

−

−

− +

−

− −

++

1 (1 )1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

) lnˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ

lnˆ ˆ


1 , 1 ,,

, 1

, ,

,, 2 , ,

, ,

, ,, ,

,X

Y

C

Yr r

W L

P Y

W L

PL

I

YI

G

YG

D M

Y

M

Y

i

i

C i

it

A

C i tA

A

C i tS i t

C

i tC

ii i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

i tM

j t

i tX

j tM

M j

jj tM i

ii t

M

A H

P P

P T T∑

αα α

φ σ αα

φφ

φφ

νν

φ Π Πτ

ττ

νν ν

ψ

=+

++

− −

−−+ −

+ −−

−

− + − −−

− +

+ − −

− −

− + ++

=

ln ˆ1

ln ˆ 1

1E ln ˆ 1 (1 )

1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ

( ) lnˆ ˆ


, , 1 , 1 , 1 ,,

, 1

2,

,

, ,

,, 2 , ,

2 ,, ,

, ,, ,

1

,Y Y YX

Y

C

Yr r

LP Y P

W L

P Y

W L

PL

I

YI

G

YG

X

YL w

D M

Y

M

Y

i t i t t i ti

i

C i

it

A

C i tA

A

C i tS i t

C

i tC

C iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

i

ii jX i t

Mj t

i tX

j tM

M j

jj tM i

ii t

M

j

N

A H

P P

T

φ σ αα

φφ

φφ

νν

ττ

ψ

−−+ −

+ −−

−

−

− +

−

− −

++

1 )1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

nˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ

ˆ1 ln ˆ 1 ln ˆ .

1 ,,

, 1

, ,

,, 2 , ,

, ,

, ,, ,

,C

Yr r

W L

PL

I

YI

G

YG

D M

Y

M

Y

i

i

C i

it

A

C i tA

A

C i tS i t

C

i tC

i i

i

i t i t

i tC

ii t

i

ii t

i

ii t

j t

j tM

M j

jj tM i

ii t

M

A H P P

P T T∑

αα α

φ σ αα

φφ

φφ

νν

φ Π Πτ

ττ

νν ν

ψ

=+

++

− −

−−+ −

+ −−

−

− + − −−

− +

+ − −

− −

− + ++

=

ln ˆ1

ln ˆ 1

1E ln ˆ 1 (1 )

1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ

( ) lnˆ ˆ


, , 1 , 1 , 1 ,,

, 1

2,

,

, ,

,, 2 , ,

2 ,, ,

, ,, ,

1

,Y Y YX

Y

C

Yr r

LP Y P

W L

P Y

W L

PL

I

YI

G

YG

X

YL w

D M

Y

M

Y

i t i t t i ti

i

C i

it

A

C i tA

A

C i tS i t

C

i tC

C iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

i

ii jX i t

Mj t

i tX

j tM

M j

jj tM i

ii t

M

j

N

A H

P T T∑

φ σ αα

φφ

φφ

νν

Π Πτ

ττ

νν ν

ψ

+− −

−−+ −

+ −−

−

− −−

− +

+ − −

− −

++

=

1

1E ln ˆ 1 (1 )

1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ

( ) lnˆ ˆ


1 , 1 ,,

, 1

,

,

, ,

,, 2 , ,

2 ,, ,

, ,, ,

1

,YX

Y

C

Yr r

P Y P

W L

P Y

W L

PL

I

YI

G

YG

X

YL w

D M

Y

M

Y

t i ti

i

C i

it

A

C i tA

A

C i tS i t

C

i tC

iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

i

ii jX i t

Mj t

i tX

j tM

M j

jj tM i

ii t

M

j

N

A H

(4.3)

Reflecting the existence of credit constraints, output also de-pends on contemporaneous, past, and expected future real profit and disposable labor income. In addition, output de-pends on contemporaneous, past, and expected future invest-ment and public domestic demand. Finally, reflecting the existence of international trade linkages, output depends on contemporaneous, past, and expected future export weighted foreign demand, as well as the export weighted average for-eign external terms of trade and the domestic external terms of trade. The response coefficients of this relationship vary across economies with the composition of their domestic de-mand, the size of their government, their labor income share, their trade openness, and their trade pattern.

Domestic demand ln ˆ,Di t depends on a weighted average

of its past and expected future values driven by a weighted average of the contemporaneous real ex ante portfolio return and short- term real market interest rate according to domes-tic demand relationship:

αα α

φ σ αα

φφ

φφ

νν

φ Π Πτ

ττ

=+

++

− −−+ −

+ −−

−

+ + − −−

+ +

− + ++

ln ˆ1

ln ˆ 1

1ln ˆ (1 )

1

1 1ˆ 1

1ˆ ln

ˆ

ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ .

, , 1 , 1 , 1 ,,

, 1

2,

,

, ,

,, 2 , ,

,D D DC

Yr r

LP Y P

W L

P Y

W L

PL

I

YI

G

YG

i t i t t i tC i

it

A

C i tA

A

C i tS i t

C

i tC

C iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

A HE E

φ σ αα

φφ

φφ

νν

ττ

τ

− −−+ −

+ −−

−

− −−

+ +

++

ln ˆ (1 )1

1 1ˆ 1

1ˆ ln

ˆ

ˆ

ˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ .

1 , 1 ,,

, 1

,

,

, ,

,, 2 , ,

,DC

Yr r

W L

P Y

W L

PL

I

YI

G

YG

tC i

it

A

C i tA

A

C i tS i t

C

i tC

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

A HE E

αα α

φ σ αα

φφ

φφ

νν

φ Π Πτ

ττ

=+

++

− −−+ −

+ −−

−

+ + − −−

+ +

− + ++

ln ˆ1

ln ˆ 1

1ln ˆ (1 )

1

1 1ˆ 1

1ˆ ln

ˆ

ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ .

, , 1 , 1 , 1 ,,

, 1

2,

,

, ,

,, 2 , ,

,D D DC

Yr r

LP Y P

W L

P Y

W L

PL

I

YI

G

YG

i t i t t i tC i

it

A

C i tA

A

C i tS i t

C

i tC

C iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

A HE E

φ σ αα

φφ

φφ

νν

ττ

−+ −

+ −−

−

−

+ +

++

)1

1 1ˆ 1

1ˆ ln

ˆ

ˆ

nˆ ˆ

ˆ1

1ˆ ( ) ln ˆ ln ˆ .

1 ,,

, 1

, ,

,, 2 , ,

,C

Yr r

W L

PL

I

YI

G

YG

C i

it

A

C i tA

A

C i tS i t

C

i tC

i i

iY

i

i t i t

i tC

ii t

i

ii t

i

ii t

A H

(4.4)

Reflecting the existence of credit constraints, domestic de-mand also depends on contemporaneous, past, and expected future real profit and disposable labor income. Finally, do-mestic demand depends on contemporaneous, past, and

expected future investment and public domestic demand. The response coefficients of this relationship vary across economies with the composition of their domestic demand, the size of their government, and their labor income share.

Consumption ln ˆ,Ci t depends on a weighted average of its

past and expected future values driven by a weighted average of the contemporaneous real ex ante portfolio return and short- term real market interest rate according to consump-tion demand relationship:

αα α

φ σ αα

φφ

φφ

φ Π Πτ

ττ

=+

++

− −−+ −

+ −−

+

+ − −−

− + +

−

ln ˆ1

ln ˆ 1

1E ln ˆ (1 )

1

1E

1ˆ 1

1

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ .

, , 1 , 1 ,

1

2,

,

, ,

,,

,C C C r r

C

YL

P Y P

W L

P Y

W L

P

i t i t t i tC

t

A

C i tA

A

C

C i

i

iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

A H

αα α

φ σ αα

φφ

φφ

νν

φ Π Πτ

ττ

=+

++

− −−+ −

+ −−

−

+

+ − −−

− + ++

−

ln ˆ1

ln ˆ 1

1E ln ˆ (1 )

1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ .

, , 1 , 1 , 1 ,,

, 11

2,

,

, ,

,,

,C C C r r

C

YL

P Y P

W L

P Y

W L

P

i t i t t i tC

t

A

C i tA

A

C i tS i t

C

i tC

C i

i

iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

A H

αα α

φ σ αα

φφ

φφ

φ Π Πτ

ττ

=+

++

− −−+ −

+ −−

+

+ − −−

− + +

−

ln ˆ1

ln ˆ 1

1E ln ˆ (1 )

1

1E

1ˆ 1

1ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ .

, , 1 , 1 , 1 ,

1

2,

,

, ,

,,

,C C C r r

C

YL

P Y P

W L

P Y

W L

P

i t i t t i tC

t

A

C i tA

A

C i tS

C i

i

iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

A H

αα α

φ σ αα

φφ

φφ

νν

φ Π Πτ

ττ

=+

++

− −−+ −

+ −−

−

+

+ − −−

− + ++

−

ln ˆ1

ln ˆ 1

1E ln ˆ (1 )

1

1E

1ˆ 1

1ˆ ln

ˆ

ˆ

( ) lnˆ

ˆ(1 ) ln

ˆ ˆ

ˆ1

1ˆ .

, , 1 , 1 , 1 ,,

, 11

2,

,

, ,

,,

,C C C r r

C

YL

P Y P

W L

P Y

W L

P

i t i t t i tC

t

A

C i tA

A

C i tS i t

C

i tC

C i

i

iS

iY

i

i tS

i tC i

i i

iY

i

i t i t

i tC

ii t

A H

(4.5)

Reflecting the existence of credit constraints, consumption also depends on contemporaneous, past, and expected fu-ture real profit and disposable labor income, where polyno-

mial in the lag operator ( ) 1 E1

112

1 = − −αα α+ +

−L L Lt . The

response coefficients of this relationship vary across econo-mies with their consumption intensity, the size of their gov-ernment, and their labor income share.

Investment ln ˆ,Ii t depends on a weighted average of its

past and expected future values driven by the contempora-neous relative shadow price of capital according to invest-ment demand relationship:

β

ββ

χ βν

=+

++

++

− +ln ˆ 11

ln ˆ1

E ln ˆ

1(1 )

ln ˆˆ

ˆ .

, , 1 , 1

,,

,

I I I

Q

P

i t i t t i t

i tI i t

i tC

(4.6)

ββ

β

χ βν

=+

++

++

− +ln ˆ 11

ln ˆ1

E ln ˆ

1(1 )

ln ˆˆ

ˆ .

, , 1 , 1

,,

,

I I I

Q

P

i t i t t i t

i tI i t

i tC

Reflecting the existence of a financial accelerator mecha-

nism, the relative shadow price of capital lnˆ ,ˆ

,

Qi t

Pi tC

depends on

its expected future value as well as the contemporaneous real ex ante portfolio return, and the contemporaneous real ex ante corporate loan rate net of the expected future loan de-fault rate, according to investment Euler equation:

β δ φ φβ θθ

κ β χ δβ

β δ φβ θθ

κ β χ δβ β

η

= − − − +−

+ − − −

+ − − +−

+ − − −

−

+

++ +ln

ˆ

ˆ E (1 )lnˆ

ˆ (1 ) ˆ1

1 (1 (1 ))( ˆ

(1 (1 ))1

1 (1 (1 )) 1ln ˆ

1

,

,

, 1

, 1, 1 , 1

,

, 1

,Q

P

Q

Pr r

u

i t

i tC t

i t

i tC i t

AC

C

R B C

i tC E

C

C

R B CK

i tK

A H

β δ φ φβ θθ

κ β χ δβ

λ δ

β δ φβ θθ

κ β χ δβ β

ητ

τ

= − − − +−

+ − − −

+ − − +−

+ − − −

−

−

+

++ + +

+ +

lnˆ

ˆ E (1 )lnˆ

ˆ (1 ) ˆ1

1 (1 (1 ))( ˆ ˆ )

(1 (1 ))1

1 (1 (1 )) 1ln ˆ 1

1ˆ .

,

,

, 1

, 1, 1 , 1

,, 1

, 1 , 1

,Q

P

Q

Pr r

u

i t

i tC t

i t

i tC i t

AC

C

R B C

i tC E Q

i tC

C

C

R B CK

i tK

ii t

A H

β δ φ φβ θθ

κ β χ δβ

β δ φβ θθ

κ β χ δβ β

ητ

= − − − +−

+ − − −

+ − − +−

+ − − −

−

−

+

++ +ln

ˆ

ˆ E (1 )lnˆ

ˆ (1 ) ˆ1

1 (1 (1 ))( ˆ

(1 (1 ))1

1 (1 (1 )) 1ln ˆ 1

1

,

,

, 1

, 1, 1 , 1

,

, 1

,Q

P

Q

Pr r

u

i t

i tC t

i t

i tC i t

AC

C

R B C

i tC E Q

C

C

R B CK

i tK

A H

β δ φ φβ θθ

κ β χ δβ

λ δ

β δ φβ θθ

κ β χ δβ β

ητ

τ

= − − − +−

+ − − −

+ − − +−

+ − − −

−

−

+

++ + +

+ +

lnˆ

ˆ E (1 )lnˆ

ˆ (1 ) ˆ1

1 (1 (1 ))( ˆ ˆ )

(1 (1 ))1

1 (1 (1 )) 1ln ˆ 1

1ˆ .

,

,

, 1

, 1, 1 , 1

,, 1

, 1 , 1

,Q

P

Q

Pr r

u

i t

i tC t

i t

i tC i t

AC

C

R B C

i tC E Q

i tC

C

C

R B CK

i tK

ii t

A H

(4.7)



Under a flexible inflation targeting regime j = 0, and the de-sired nominal policy interest rate responds to contemporane-ous consumption price inflation and output. Under a managed exchange rate regime j = 1, and it also responds to the contemporaneous real effective exchange rate. Under a fixed exchange rate regime j = 2, and the nominal policy in-terest rate instead tracks the contemporaneous nominal pol-icy interest rate for the economy that issues the anchor currency one for one, while responding to the contempora-neous corresponding nominal bilateral exchange rate. For economies belonging to a currency union, the target vari-ables entering into their common monetary policy rule are expressed as output weighted averages across union members. The real policy interest rate ˆ,ri t

P satisfies ˆ ˆ E ˆ, , , 1r ii tP

i tP

t i tCπ= − + .

The short- term nominal market interest rate ˆ,ii tS depends on

the contemporaneous nominal policy interest rate and the net foreign asset ratio according to the money market relationship,

ˆ ˆˆ

ˆ ,, ,, 1

, ,,i i

AP Yi t

Si tP i i t

i tY

i ti tiSζ υ= − ++

(4.13)

where credit risk premium shock ˆ ,i tiSυ satisfies dynamic factor

process ˆ ˆ (1 ) ˆ, ,1 ,w wi ti

kM

jM

j ti

jN

kM

iM

i tiS S S∑υ λ ν λ ν= + −= . The inten-

sity of international money market contagion varies across economies, with k = 0 for low- debt contagion economies, k = 1 for medium- debt contagion economies, and k = 2 for high- debt contagion economies, where 0 1 2

M M Mλ λ λ< < . For econ-omies belonging to a currency block, the ratio of national financial wealth to nominal output is expressed as an output weighted average across block members. The short- term real market interest rate ˆ,ri t

S satisfies ˆ ˆ E ˆ, , , 1r ii tS

i tS

t i tCπ= − + .

The long- term nominal market interest rate ˆ,ii tL depends

on a weighted average of its expected future value and the contemporaneous short- term nominal market interest rate according to bond market relationship,

β β υ= + − −+ˆ E ˆ (1 )( ˆ ln ˆ ),, , 1 , ,i i ii t

Lt i t

Li tS

i tB (4.14)

where duration risk premium shock ln ˆ ,i tBυ satisfies dynamic

factor process ln ˆ ln ˆ (1 ) ln ˆ, ,1 ,w wi tB

kB

jB

j tB

jN

kB

iB

i tB∑υ λ ν λ ν= + −= .

The intensity of international bond market contagion varies across economies, with k = 0 for low- debt contagion econo-mies, k = 1 for medium- debt contagion economies, and k = 2 for high- debt contagion economies, where 0 1 2

B B Bλ λ λ< < . The long- term real market interest rate ˆ,ri t

L satisfies the same bond market relationship, driven by the contemporaneous short- term real market interest rate.

The price of equity ln ˆ,Vi tS depends on its expected future

value driven by expected future net profits and the contem-poraneous short- term nominal market interest rate accord-ing to the stock market relationship,

V V ii tS

t i tS

t i tS

i tS

i tSln ˆ E ln ˆ (1 )E ln ˆ ( ˆ ln ˆ ),, , 1 , 1 , ,β β Π υ= + − − −+ + (4.15)

where equity risk premium shock ln ˆ ,i tSυ satisfies dynamic

factor process ln ˆ ln ˆ (1 ) ln ˆ, ,1 ,w wi tS

kS

jS

j tS

jN

kS

iS

i tS∑υ λ ν λ ν= + −= .

The intensity of international stock market contagion varies

The relative shadow price of capital also depends on the ex-pected future capital utilization and tax rates. Auxiliary pa-rameter λQ is theoretically predicted to equal one, and satisfies λQ ≥ 0. The capital utilization rate ln ˆ ,ui t

K depends on the contemporaneous real wage according to capital utiliza-tion relationship:

η

= −

ln ˆ 1

lnˆ

ˆ lnˆ ˆ

ˆ .,,

,

, ,

,

uW

P

u K

Li tK

K

i t

i tC

i tK

i t

i t

(4.8)

The capital utilization rate also depends on the contempora-neous deviation of utilized capital from employment. The capital stock ln ˆ

, 1K i t + satisfies ln ˆ (1 ) ln ˆ ln( ˆ ˆ ), 1 , , ,K K Ii t i t i tI

i tδ δ ν= − ++) ln ˆ ln( ˆ ˆ ), , ,Ii t i t

Ii tδ δ ν+ .

Exports ln ˆ,Xi t depend on contemporaneous export

weighted foreign demand as well as the export weighted av-erage foreign external terms of trade, according to export demand relationship:

∑ ν νψ= − −

=

ln ˆ lnˆ

ˆ ˆ1 ln ˆ ., ,

,

, ,,

1

X wD M

Yi t i jX j t

i tX

j tM

M j

jj tM

j

N

(4.9)

The response coefficients of this relationship vary across economies with their trade pattern and the trade openness of their trading partners. Imports ln ˆ

,Mi t depend on contempo-raneous domestic demand, as well as the domestic external terms of trade, according to import demand relationship:

ν

ψ= − −

ln ˆ lnˆ

ˆ1 ln ˆ .,

,

,,M

D M

Yi ti t

i tM

M i

ii t

M (4.10)

The response coefficients of this relationship vary across economies with their trade openness.

The nominal ex ante portfolio return Ε +it i tA A Hˆ, 1

, depends on the contemporaneous short- term nominal market interest rate according to return function:

∑ ∑υνν

υνν

= − −

− −

ε

ε

ε

ε+= =

E ˆ ˆ ln ˆ lnˆ

ˆln ˆ ln

ˆ

ˆ., 1 ,

,

, , , ,

,

,1, , ,

,

,1

,i iB

Aw

B

B

S

Awt i t

Ai tS i

L H

iA H i j

Bj tB i

H

iL H

i t

j tj

NiH

iA H i j

Sj tS i t

j tj

NA H

∑ ∑υνν

υνν

−

− −

ε

ε

ε

ε= =

ln ˆ lnˆ

ˆln ˆ ln

ˆ

ˆ., , ,

,

,1, , ,

,

,1

wB

B

S

Awi j

Bj tB i

H

iL H

i t

j t

NiH

iA H i j

Sj tS i t

j tj

N

(4.11)

Reflecting the existence of internal and external macro- financial linkages, the nominal ex ante portfolio return also depends on contemporaneous domestic and foreign duration risk premium, equity risk premium, and currency risk pre-mium shocks. The response coefficients of this relationship vary across economies with their domestic and foreign money, bond, and stock market exposures. The real ex ante portfolio return Ε +rt i t

A A Hˆ, 1, satisfies π= −+ + +r it i t

At i t

At i t

CA H A HE ˆ E ˆ E ˆ, 1 , 1 , 1, , .

The nominal policy interest rate ˆ,ii tP depends on a

weighted average of its past and desired values according to monetary policy rule:

Q

E

ρ ρ ξ π ξ ξξ ξ ν

= + − + ++ + +

π

ε−

ˆ ˆ (1 )( ˆ ln ˆ ln ˆ

ˆ ln ˆ ) ˆ ., , 1 , ,

Q,

, , , ,

i i Y

ii tP

ji

i tP

ji

j i tC

jY

i t j i t

ji

k tP

j i k t i tiP

Q

E

ρ ρ ξ π ξ ξξ ξ ν

= + − + ++ + +

π

ε−

ˆ ˆ (1 )( ˆ ln ˆ ln ˆ

ˆ ln ˆ ) ˆ ., , 1 , ,

Q,

, , , ,

i i Y

ii tP

ji

i tP

ji

j i tC

jY

i t j i t

ji

k tP

j i k t i tiP (4.12)


Francis Vitek 67

foreign loan default rates according to credit- loss rate function:

ˆ ˆ .,,

, ,1

wi tC E

i jC

j tC

j

N

∑δ δ==

(4.19)

The real ex ante corporate loan rate E ˆ, 1,+rt i t

C E satisfies E ˆ E ˆ E ˆ, 1

,, 1

,, 1π= −+ + +r it i t

C Et i t

C Et i t

C .The nominal bank lending interest rate ˆ

,ii tC depends on a

weighted average of its past and expected future values driven by the deviation of the past short- term nominal mar-ket interest rate from the contemporaneous nominal bank lending interest rate net of the contemporaneous credit- loss rate according to lending rate Phillips curve:

ˆ 11

ˆ1

E ˆ (1 )(1 )(1 )

ˆ ( ˆ ˆ )

1 (1 )1 (1 (1 ))

( ˆ ˆ ) ( ˆ ˆ )1

, , 1 , 1 , 1 , ,,

, , , ,

i i i i i

i

i tC

i tC

t i tC

C C

C i tS

i tC

i tC E

B C

R B CC

i t i tR

i tR R

i tS

C

ββ

βω ω βω β

δ

β χ δκ β χ δ

η κ κ κ κθ

=+

++

+ − −+

− −

− − −+ − −

− − − −

− + −

−

ˆ 11

ˆ1

E ˆ (1 )(1 )(1 )

ˆ ( ˆ ˆ )

1 (1 )1 (1 (1 ))

( ˆ ˆ ) ( ˆ ˆ )1

1ln ˆ .

, , 1 , 1 , 1 , ,,

, , , , 1 ,

i i i i i

i

i tC

i tC

t i tC

C C

C i tS

i tC

i tC E

B C

R B CC

i t i tR

i tR R

i tS

C i tC

ββ

βω ω βω β

δ

β χ δκ β χ δ

η κ κ κ κθ

θ

=+

++

+ − −+

− −

− − −+ − −

− − − −−

− + −

−

ˆ 11

ˆ1

E ˆ (1 )(1 )(1 )

ˆ ( ˆ ˆ )

1 (1 )1 (1 (1 ))

( ˆ ˆ ) ( ˆ ˆ )1

1ln

, , 1 , 1 , 1 , ,,

, , , ,

i i i i i

i

i tC

i tC

t i tC

C C

C i tS

i tC

i tC E

B C

R B CC

i t i tR

i tR R

i tS

C

ββ

βω ω βω β

δ

β χ δκ β χ δ

η κ κ κ κθ

=+

++

+ − −+

− −

− − −+ − −

− − − −−

− + −

−

ˆ 11

ˆ1

E ˆ (1 )(1 )(1 )

ˆ ( ˆ ˆ )

1 (1 )1 (1 (1 ))

( ˆ ˆ ) ( ˆ ˆ )1

1ln ˆ .

, , 1 , 1 , 1 , ,,

, , , , 1 ,

i i i i i

i

i tC

i tC

t i tC

C C

C i tS

i tC

i tC E

B C

R B CC

i t i tR

i tR R

i tS

C i tC

ββ

βω ω βω β

δ

β χ δκ β χ δ

η κ κ κ κθ

θ

=+

++

+ − −+

− −

− − −+ − −

− − − −−

− + −

−

(4.20)

Reflecting the existence of a regulatory capital requirement, the nominal bank lending interest rate also depends on the past deviation of the bank capital ratio from its required value, as well as the past deviation of the regulatory bank capital ratio from its funding cost. The real bank lending interest rate ˆ,ri t

C satisfies ˆ ˆ E ˆ, , , 1r ii tC

i tC

t i tCπ= − + .

The money stock ln ˆ, 1Mi tS

+ depends on contemporaneous bank credit and the bank capital stock according to bank balance sheet identity:

ln ˆ (1 )ln ˆ ln ˆ ., 1,

, 1 , 1B M Ki tC B R

i tS R

i tBκ κ= − ++ + + (4.21)

The bank capital ratio ˆ , 1i tκ + satisfies ˆ (ln ˆ ln, 1 , 1i tR

i tBκ κ= −+ +

ˆ (ln ˆ ln ˆ ), 1 , 1 , 1,K Bi t

Ri tB

i tC Bκ κ= −+ + + . Retained earnings ln ˆ

,Ii tB depends on a

weighted average of its past and expected future values driven by the contemporaneous shadow price of bank capital according to retained earnings relationship:

ln ˆ 11

ln ˆ1

E ln ˆ 1(1 )

ln ˆ ., , 1 , 1 ,I I I Qi tB

i tB

t i tB

C i tB

ββ

β χ β=

++

++

+− + (4.22)

The shadow price of bank capital ln ˆ,Qi tB depends on its ex-

pected future value net of the expected future credit- loss rate, as well as the contemporaneous short- term nominal market interest rate, according to retained earnings Euler equation:

ln ˆ E (1 )(ln ˆ ˆ ) ˆ (1 (1 )) ( ˆ ˆ, , 1 , 1,

, ,Q Q ii tB

tB C

i tB

i tC E

i tS B C

C

R i tβ χ δ δ β χ δ ηκ

κ κ= − − − + − − −

+ +

ln ˆ E (1 )(ln ˆ ˆ ) ˆ (1 (1 )) ( ˆ ˆ ) ., , 1 , 1,

, , 1 , 1Q Q ii tB

tB C

i tB

i tC E

i tS B C

C

R i t i tRβ χ δ δ β χ δ η

κκ κ= − − − + − − −

+ + + +

(4.23)

Reflecting the existence of a regulatory capital requirement, the shadow price of bank capital also depends on the

across economies, with k = 0 for low- equity contagion econ-omies, k = 1 for medium- equity contagion economies, and k = 2 for high- equity contagion economies, where

0 1 2S S Sλ λ λ< < .

Real net profits Π

lnˆ

,ˆ

,

i tS

Pi tY depends on contemporaneous

output, the labor income share, and the tax rate, as well as the deviation of investment from output and the terms of trade, according to profit function:

PY

P YW LP Y

W L

P Y

W LP Y

IY

B

P Y

iP Y

P Y

B

P Y

I

Y

XY

i tS

i tY i t

iS

iY

ii

i i

iY

i

i t i t

i tY

i t

i i

iY

ii t

i

i

i tC F

i tY

i t

C

C

R B C

i tC E

i tC i t

Yi t

i tY

i t

i tC F

i tY

i t

i t

i t

i

i

i tX

i tM

lnˆ

ˆ ln ˆ (1 ) lnˆ ˆ

ˆ ˆ 1 ˆ lnˆ

ˆ ˆ

11 (1 (1 )) ˆ ˆ ln

ˆ ˆ

ˆ ˆ lnˆ

ˆ ˆ lnˆ

ˆ lnˆ

ˆ .

,

,,

1

, ,

, ,,

, 1,

, ,

,,

,, ,

, 1 , 1

,,

, 1 , 1

,

,

,

,

Π Π τ τ λ φδ

θθ

κ β χ δβ

δ

= −

− + −

−

−−

+ − − − −

+

− −

Π

−+

− − − −Y

P YW LP Y

W L

P Y

W LP Y

IY

B

P Y

iP Y

P Y

B

P Y

I

Y

XY

i ti

S

iY

ii

i i

iY

i

i t i t

i tY

i t

i i

iY

ii t

i

i

i tC F

i tY

i t

C

C

R B C

i tC E

i tC i t

Yi t

i tY

i t

i tC F

i tY

i t

i t

i t

i

i

i tX

i tM

ˆ (1 ) lnˆ ˆ

ˆ ˆ 1 ˆ lnˆ

ˆ ˆ

11 (1 (1 )) ˆ ˆ ln

ˆ ˆ

ˆ ˆ lnˆ

ˆ ˆ lnˆ

ˆ lnˆ

ˆ .

,

1

, ,

, ,,

, 1,

, ,

,,

,, ,

, 1 , 1

,,

, 1 , 1

,

,

,

,

Π τ τ λ φδ

θθ

κ β χ δβ

δ

− + −

−

−−

+ − − − −

+

− −

Π

−+

− − − −

PY

P YW LP Y

W L

P Y

W LP Y

IY

B

P Y

iP Y

P Y

B

P Y

I

Y

XY

i tS

i tY i t

iS

iY

ii

i i

iY

i

i t i t

i tY

i t

i i

iY

ii t

i

i

i tC F

i tY

i t

C

C

R B C

i tC E

i tC i t

Yi t

i tY

i t

i tC F

i tY

i t

i t

i t

i

i

i tX

i tM

lnˆ

ˆ ln ˆ (1 ) lnˆ ˆ

ˆ ˆ 1 ˆ lnˆ

ˆ ˆ

11 (1 (1 )) ˆ ˆ ln

ˆ ˆ

ˆ ˆ lnˆ

ˆ ˆ lnˆ

ˆ lnˆ

ˆ .

,

,,

1

, ,

, ,,

, 1,

, ,

,,

,, ,

, 1 , 1

,,

, 1 , 1

,

,

,

,

Π Π τ τ λ φδ

θθ

κ β χ δβ

δ

= −

− + −

−

−−

+ − − − −

+

− −

Π

−+

− − − −

YP Y

W LP Y

W L

P Y

W LP Y

IY

B

P Y

iP Y

P Y

B

P Y

I

Y

XY

i tS

Y i ti

S

iY

ii

i i

iY

i

i t i t

i tY

i t

i i

iY

ii t

i

i

i tC F

i tY

i t

C

C

R B C

i tC E

i tC i t

Yi t

i tY

i t

i tC F

i tY

i t

i t

i t

i

i

i tX

i tM

ln ˆ (1 ) lnˆ ˆ

ˆ ˆ 1 ˆ lnˆ

ˆ ˆ

11 (1 (1 )) ˆ ˆ ln

ˆ ˆ

ˆ ˆ lnˆ

ˆ ˆ lnˆ

ˆ lnˆ

ˆ .

,

,,

1

, ,

, ,,

, 1,

, ,

,,

,, ,

, 1 , 1

,,

, 1 , 1

,

,

,

,

Π τ τ λ φδ

θθ

κ β χ δβ

δ

= −

− + −

−

−−

+ − − − −

+

− −

Π

−+

− − − −

W LP Y

IY

B

P Y

P Y

P Y

B

P Y

I

Y

XY

i i

iY

ii t

i

i

i tC F

i tY

i t

i tC i t

Yi t

i tY

i t

i tC F

i tY

i t

i t

i t

i

i

i tX

i tM

1 ˆ lnˆ

ˆ ˆ

ˆ lnˆ ˆ

ˆ ˆ lnˆ

ˆ ˆ lnˆ

ˆ lnˆ

ˆ .

,, 1

,

, ,

,,

, ,

, 1 , 1

,,

, 1 , 1

,

,

,

,

λ φδ

δ

−

−

− −

+

− −

Π +

− − − −

(4.16)

Reflecting the existence of a financial accelerator mecha-nism, real net profits also depend on the contemporaneous and past nonfinancial corporate debt ratio, as well as the contemporaneous nominal corporate loan rate net of the contemporaneous loan default rate and nominal output growth rate. Auxiliary parameter λΠ is theoretically pre-dicted to equal one, and satisfies 0λ ≥Π . The response coef-ficients of this relationship vary across economies with the size of their government, their labor income share, their in-vestment intensity, and their trade openness.

Reflecting the existence of international bank lending linkages, bank credit ln ˆ

, 1,Bi t

C B+ depends on the contempora-

neous bank lending weighted average of domestic currency denominated domestic and foreign nonfinancial corporate debt according to bank credit- demand function:

∑=++

=ln ˆ ln

ˆ

ˆ., 1

,,

, 1,

, ,1

B wB

i tC B

i jC j t

C F

j i tj

N

(4.17)

Nonfinancial corporate debt satisfies ln ˆ ln ˆ ln ˆ, 1

,, , 1B P Ki t

C Fi tC

i t= ++ +ˆ ln ˆ ln ˆ

, 1,

, , 1B P Ki tC F

i tC

i t= ++ + . Furthermore, the nominal corporate loan rate ˆ

,,ii t

C E depends on the nonfinancial corporate- borrowing weighted average of past domestic and foreign nominal bank lending interest rates, adjusted for contemporaneous changes in nominal bilateral exchange rates, according to:

∑= +

−

−=

ˆ ˆ lnˆ

ˆ.,

,, , 1

, ,

, , 11

i w ii tC E

i jF

j tC i j t

i j tj

N

(4.18)

Finally, the credit- loss rate ˆ,

,i tC Eδ depends on the bank

lending weighted average of contemporaneous domestic and



The internal terms of trade ln ˆ,i tX depends on the con-

temporaneous relative domestic currency- denominated prices of energy and nonenergy commodities according to internal terms of trade function:

T

E∑ ∑= −

=

−

=ln ˆ 1 ln

ˆ ˆ

ˆ.,

,

1

*1

, , *, ,

,1

*X

Y

X

X

X

X

P

Pi t

X i

i

i k

ik

Mi k

i

i i t k tY

i tY

k

M

(4.28)

The response coefficients of this relationship vary across economies with their trade openness and commodity export intensities.

The change in the external terms of trade ln ˆ,i tM depends

on a linear combination of its past and expected future val-ues driven by the contemporaneous deviation of the import weighted average real bilateral exchange rate from the exter-nal terms of trade according to import price Phillips curve:

T T T

Q

TT T

P T P E

∑

∑

γ µγ β µ

βγ β µ

ω ω βω γ β µ θ

π γ βγ β µ

µ

∆ =−

+ −∆ +

+ −∆

+− −( + −

+ + −

−−

− − ∆

++

+ −

− +

=

=

ln ˆ (1 )

1 (1 )ln ˆ

1 (1 )E ln ˆ

(1 )(1 )

1 (1 ))ln

ˆ

ˆln ˆ 1 ln ˆ 1

( ) ˆ ln ˆ (1 )

1 (1 )( ) ln( ˆ ˆ

, , 1 , 1

,, ,

,

, ,1

4 , , 5 , , *, ,1

*

wX

Y

X

Y

LX

YL P

i tM

MiM

MiM i t

MM

iM t i t

M

M M

M MiM i j

M i j t

i tM

i

ii t

X j

jj t

X

j

N

M

i tY i

ii t

XM

MiM i k

Mi i t k t

Y

k

M

T T T

Q

TT T

P T P E

∑

∑

γ µγ β µ

βγ β µ


π γ βγ β µ

µ

∆ =−

+ −∆ +

+ −∆

+− −( + −

+ + −

−−

− − ∆

++

+ −

− +

=

=

ln ˆ (1 )

1 (1 )ln ˆ

1 (1 )E ln ˆ

(1 )(1 )

1 (1 ))ln

ˆ

ˆln ˆ 1 ln ˆ 1

1

( ) ˆ ln ˆ (1 )

1 (1 )( ) ln( ˆ ˆ

, , 1 , 1

,, ,

,

, ,1

4 , , 5 , , *, ,1

*

wX

Y

X

Y

LX

YL P

i tM

MiM

MiM i t

MM

iM t i t

M

M M

M MiM i j

M i j t

i tM

i

ii t

X j

jj t

X

j

N

M

i tY i

ii t

XM

MiM i k

Mi i t k t

Y

k

M

T T T

Q

TT T

P T P E

∑

∑

γ µγ β µ

βγ β µ


θ

π γ βγ β µ

µ

∆ =−

+ −∆ +

+ −∆

+− −( + −

+ + −

−−

− − ∆

++

+ −

− +

=

=

ln ˆ (1 )

1 (1 )ln ˆ

1 (1 )E ln ˆ

(1 )(1 )

1 (1 ))ln

ˆ

ˆln ˆ 1 ln ˆ 1

1ln ˆ

( ) ˆ ln ˆ (1 )

1 (1 )( ) ln( ˆ ˆ ).

, , 1 , 1

,, ,

,

, ,1

,

4 , , 5 , , *, ,1

*

wX

Y

X

Y

LX

YL P

i tM

MiM

MiM i t

MM

iM t i t

M

M M

M MiM i j

M i j t

i tM

i

ii t

X j

jj t

X

j

N

M i tM

i tY i

ii t

XM

MiM i k

Mi i t k t

Y

k

M

T T T

Q

TT T

P T P E

∑

∑

γ µγ β µ

βγ β µ


θ

π γ βγ β µ

µ

∆ =−

+ −∆ +

+ −∆

+− −( + −

+ + −

−−

− − ∆

++

+ −

− +

=

=

ln ˆ (1 )

1 (1 )ln ˆ

1 (1 )E ln ˆ

(1 )(1 )

1 (1 ))ln

ˆ

ˆln ˆ 1 ln ˆ 1

1ln ˆ

( ) ˆ ln ˆ (1 )

1 (1 )( ) ln( ˆ ˆ ).

, , 1 , 1

,, ,

,

, ,1

,

4 , , 5 , , *, ,1

*

wX

Y

X

Y

LX

YL P

i tM

MiM

MiM i t

MM

iM t i t

M

M M

M MiM i j

M i j t

i tM

i

ii t

X j

jj t

X

j

N

M i tM

i tY i

ii t

XM

MiM i k

Mi i t k t

Y

k

M

T T T

Q

TT T

P T P E

∑

∑

γ µγ β µ

βγ β µ


θ

π γ βγ β µ

µ

∆ =−

+ −∆ +

+ −∆

+− −( + −

+ + −

−−

− − ∆

++

+ −

− +

=

=

ln ˆ (1 )

1 (1 )ln ˆ

1 (1 )E ln ˆ

(1 )(1 )

1 (1 ))ln

ˆ

ˆln ˆ 1 ln ˆ 1

1ln ˆ

( ) ˆ ln ˆ (1 )

1 (1 )( ) ln( ˆ ˆ ).

, , 1 , 1

,, ,

,

, ,1

,

4 , , 5 , , *, ,1

*

wX

Y

X

Y

LX

YL P

i tM

MiM

MiM i t

MM

iM t i t

M

M M

M MiM i j

M i j t

i tM

i

ii t

X j

jj t

X

j

N

M i tM

i tY i

ii t

XM

MiM i k

Mi i t k t

Y

k

M

The change in the external terms of trade also depends on the contemporaneous domestic and import weighted average for-eign internal terms of trade. In addition, the change in the external terms of trade depends on contemporaneous, past, and expected future output price inflation and the change in the internal terms of trade, where polynomial in the lag op-

erator ( ) 1 E(1 )

1 (1 ) 1 (1 )41 = − −

γ µ

γ β µβ

γ β µ

−

+ − + −−L L L

MiM

MiM M

iM t . Fi-

nally, the change in the external terms of trade depends on the contemporaneous, past, and expected future domestic currency- denominated prices of energy and nonenergy com-modities. The response coefficients of this relationship vary across economies with their trade openness, their trade pat-tern, and their commodity import intensities.

Public domestic demand ln ˆ,Gi t depends on a weighted

average of its past and desired values according to fiscal ex-penditure rule:

ln ˆ ln ˆ (1 )ˆ

ˆ ., , 1

1

, 1

, ,

1

,G GG

Y

A

P Y

G

Yi t G i t GG i

i

i tG

i tY

i t

i

ii tGρ ρ ζ ν= + −

+

−

−+

−

(4.30)

Desired public domestic demand responds to the contempo-raneous net government asset ratio. The tax rate ˆ ,i tτ depends on a weighted average of its past and desired values accord-ing to fiscal revenue rule:

ˆ ˆ (1 )ˆ

ˆ ., , 1, 1

, ,,

A

P Yi t i ti tG

i tY

i ti tTτ ρ τ ρ ζ ν= − − +τ τ

τ−

+ (4.31)

(4.29)

contemporaneous deviation of the bank capital ratio from its required value. The bank capital stock ln ˆ

, 1K i tB

+ satisfies ln ˆ (1 ) ln ˆ ln ˆ ˆ

, 1 , , ,,K K Ii t

B B Ci tB B C

i tB B

i tC Eχ δ χ δ χ δ= − + −+ .

The real wage lnˆ ,ˆ

,

Wi t

Pi tC

depends on a weighted average of its

past and expected future values driven by the contemporane-ous unemployment rate according to wage Phillips curve:

ββ

βω ω βω β η θ

θ γ ββ

π

=+

++

−− −

++

−

−++

−

−

+

+

lnˆ

ˆ1

1ln

ˆ

ˆ 1E ln

ˆ

ˆ

(1 )(1 )

(1 )

1ˆ

1

1ln ˆ 1

1( ) ˆ .

,

,

, 1

, 1

, 1

, 1

, , 3 ,

W

P

W

P

W

P

u L

i t

i tC

i t

i tC t

i t

i tC

L L

L i tL

L i tL

L

i tC

ββ

βω ω βω β η θ

θ γ ββ

π

=+

++

−− −

++

−

−++

−

−

+

+

lnˆ

ˆ1

1ln

ˆ

ˆ 1E ln

ˆ

ˆ

(1 )(1 )

(1 )

1ˆ

1

1ln ˆ 1

1( ) ˆ .

,

,

, 1

, 1

, 1

, 1

, , 3 ,

W

P

W

P

W

P

u L

i t

i tC

i t

i tC t

i t

i tC

L L

L i tL

L i tL

L

i tC

βββ

η θθ γ β

βπ

++

++

−

−++

−

−

+

+1E ln

ˆ

ˆ

)(1 )

(1 )

1ˆ

1

1ln ˆ 1

1( ) ˆ .

, 1

, 1

, 1

, 1

, , 3 ,

W

P

u L

C ti t

i tC

L L

L i tL

L i tL

L

i tC (4.24)

The real wage also depends on contemporaneous, past, and expected future consumption price inflation, where the polynomial in the lag operator ( ) 1 E

1 131 = − −γ

γ ββγ β+ +

−L L LL

L L t

1 E1 1

1= − −γγ β

βγ β+ +

−LL

L L t . The unemployment rate ˆ ,ui tL satisfies

ˆ ln ˆ ln ˆ, , ,u N Li tL

i t i t= − .

The labor force ln ˆ,Ni t depends on contemporaneous em-

ployment and the after- tax real wage according to the labor supply relationship:

ln ˆ ln ˆ lnˆ

ˆ1

1ˆ ln ˆ ., ,

,

,, ,N L

W

Pi t i t

i t

i tC

ii t i t

Lηι

ητ

τ ν= + −−

−

(4.25)

Employment ln ˆ,Li t depends on contemporaneous output and

the utilized capital stock according to production function:

∑ φ θθ

θθ

= − −−

+−=

ln ˆ 11

ln( ˆ ˆ )1

ln( ˆ ˆ ).,,

1

*

, , , ,YX

Y

X

X

W L

P Yu K

W L

P YLi t

i

i

i k

ikF

k

M Y

Y

i i

iY

ii tK

i t

Y

Y

i i

iY

ii t i t

φ θθ

θθ

−−

+−1

ln( ˆ ˆ )1

ln( ˆ ˆ ).,

1

*

, , , ,

X

X

W L

P Yu K

W L

P YLi k

ikF

Y

Y

i i

iY

ii tK

i t

Y

Y

i i

iY

ii t i t (4.26)

The response coefficients of this relationship vary across economies with their labor income share, their trade open-ness, and their commodity export intensities.

The nominal bilateral exchange rate ln ˆ, *,i i t depends on

its expected future value driven by the contemporaneous short- term nominal market interest rate differential accord-ing to foreign exchange market relationship:

νν

= − − +

+ln ˆ E ln ˆ ( ˆ ˆ ) ln

ˆ

ˆ., *, , *, 1 , *,

,

*,

i ii i t t i i t i tS

i tS i t

i t

(4.27)

For economies belonging to a currency union, the variables entering into their common foreign exchange market rela-tionship are expressed as output weighted averages across union members. The real bilateral exchange rate ln ˆ

, *,i i t sat-isfies ln ˆ ln ˆ ln ˆ ln ˆ

, *, , *, *, ,Q E= + −P Pi i t i i t i tY

i tY .1

1 The nominal effective exchange rate, ln ˆ,i t , satisfies ln ˆ ln ˆ ln ˆ

, , *, , , *,1 ∑= −

=wi t i i t i j

Tj i tj

N

ln ˆ ln ˆ ln ˆ, , *, , , *,1 ∑= −

=wi t i i t i j

Tj i tj

N, while the real effective exchange rate, ln ˆ

,i t satis-

fies ∑= −=

ln ˆ ln ˆ ln ˆ, , *, , , *,1

wi t i i t i jT

j i tj

N.


Francis Vitek 69

growth rate of nominal output and the fiscal balance ratio according to:

ˆ 1

1

ˆln

ˆ ˆ

ˆ ˆ ., 1

, ,

,

, 1 , 1

, ,

, 1 , 1

,

, ,

A

P Y g

A

P Y

A

P Y

P Y

P Y

FB

P Yi tG

i tY

i t

i tG

i tY

i t

iG

iY

i

i tY

i t

i tY

i t

i t

i tY

i t

�=

+−

++

− − − −

(4.36)

Finally, the effective long- term nominal market interest rate ˆ

,,ii t

L E depends on a weighted average of its past value and the contemporaneous long- term nominal market interest rate according to ˆ ˆ (1 )ˆ,

,, 1

,,i i ii t

L E Gi tL E G

i tLχ χ= + −− . The linearization

of these relationships accounts for long- term balanced growth at nominal rate g. Their response coefficients vary across economies with their public financial wealth, the size of their government, and their trade openness.

The current account balance ratio ,

*, , , ,

�

CAi t

i i t Pi tYi tY depends on

the contemporaneous quotation currency- denominated world money market portfolio return, as well as the past net foreign asset ratio, and the contemporaneous growth rate of world nominal output and the trade balance ratio, according to national dynamic budget constraint:

�

∑β

β=+

+

+ − −

−−= − −

1 1

1ˆ ln

ˆ

ˆ(1 )

ˆ,

*, , , ,, 1

*, ,

*, , 11

,

, 1 , 1

CA

P Y g

A

P Yw i

A

P Y

Ai t

i i t i tY

i t

i

iY

ijM

j tS i j t

i j tj

Ni t

i tY

i t

i

iY

� �

∑β

β=+

+

+ − −

+−−= − − − −

1 1

1ˆ ln

ˆ

ˆ(1 )

ˆln

ˆ ˆ

ˆ ˆ.

,

*, , , ,, 1

*, ,

*, , 11

,

, 1 , 1 1 1

,

*, , , ,

CA

P Y g

A

P Yw i

A

P Y

A

P Y

P Y

P Y

TB

P Yi t

i i t i tY

i t

i

iY

ijM

j tS i j t

i j tj

Ni t

i tY

i t

i

iY

i

tY

t

tY

t

i t

i i t i tY

i t

� �

∑β

β=+

+

+ − −

+−−= − − − −

1 1

1ˆ ln

ˆ

ˆ(1 )

ˆln

ˆ ˆ

ˆ ˆ.

,

*, , , ,, 1

*, ,

*, , 11

,

, 1 , 1 1 1

,

*, , , ,

CA

P Y g

A

P Yw i

A

P Y

A

P Y

P Y

P Y

TB

P Yi t

i i t i tY

i t

i

iY

ijM

j tS i j t

i j tj

Ni t

i tY

i t

i

iY

i

tY

t

tY

t

i t

i i t i tY

i t

(4.37)

Furthermore, the trade balance ratio ,E *, , , ,

TBi t

i i t Pi tY Yi t

� depends

on the contemporaneous deviation of exports from imports and the terms of trade according to:

�

E

T

T= ln

ˆ ˆ

ˆ ˆ.

,

*, , , ,

, ,

, ,

TB

P Y

X

Y

X

M

i t

i i t i tY

i t

i

i

i tX

i t

i tM

i t

(4.38)

Finally, the net foreign asset ratio ˆ , 1

, ,

+Ai tPi tYi tY

depends on its past

value, as well as the contemporaneous growth rate of world nominal output and the current account balance ratio ac-cording to:

�

=

+−

++

− − − −

ˆ 1

1

ˆln

ˆ ˆ

ˆ ˆ., 1

, ,

,

, 1 , 1 1 1

,

*, , , ,

A

P Y g

A

P Y

A

P Y

P Y

P Y

CA

P Yi t

i tY

i t

i t

i tY

i t

i

iY

i

tY

t

tY

t

i t

i i t i tY

i t

(4.39)

The linearization of these relationships accounts for long- term balanced growth at nominal rate g. Their response coef-ficients vary across economies with their national financial wealth and their trade openness.

The price of commodities ln ˆ,Pk t

Y depends on a weighted average of its past and expected future values driven by the contemporaneous world output weighted average labor income share, output, and the relative domestic currency- denominated

The desired tax rate responds to the contemporaneous net government asset ratio. The response coefficients of the for-mer relationship vary across economies with the size of their government.

The regulatory bank capital ratio ˆ , 1i tRκ + depends on a

weighted average of its past and desired values according to macroprudential policy rule:

ˆ ˆ (1 ) lnˆ

ˆ ˆ E ˆ ˆ ) ˆ ., 1 ,,

,, 1

,

, ,

,, 1 , ,

,BP Y

B

P Yi ii t

Ri tR B i

C B

iY

i

i tC B

i tY

i t

it i t

Ai tS

i tA H(κ ρ κ ρ ζ ζ ν= + − − −

+κ κ

κ κ κ+

++

(1 ) lnˆ

ˆ ˆ E ˆ ˆ ) ˆ ., 1 ,,

,, 1

,

, ,

,, 1 , ,

,BP Y

B

P Yi iR

i tR B i

C B

iY

i

i tC B

i tY

i t

it i t

Ai tS

i tA H(κ ρ ζ ζ ν= + − − −

+κ κ

κ κ κ+

++ (4.32)

The desired regulatory bank capital ratio responds to the contemporaneous bank credit ratio, as well as the contem-poraneous expected excess portfolio return. The loan de-fault rate ˆ

,i tCδ depends on a weighted average of its past and

attractor values according to default rate relationship:

ˆ ˆ (1 ) lnˆ

ˆ ˆ E ˆ ˆ ) ˆ ., , 1,

,, 1

,

, ,

,, 1 , ,

,BP Y

B

P Yi ii t

Ci tC B i

C F

iY

i

i tC F

i tY

i t

it i t

Ai tS

i tA H(δ ρ δ ρ ζ ζ ν= + − + −

+δ δ

δ δ δ−

++

(1 ) lnˆ

ˆ ˆ E ˆ ˆ ) ˆ .1,

,, 1

,

, ,

,, 1 , ,

,BP Y

B

P Yi ii t

C B iC F

iY

i

i tC F

i tY

i t

it i t

Ai tS

i tA H(δ ρ ζ ζ ν= + − + −

+δ δ

δ δ δ−

++ (4.33)

The attractor loan default rate responds to the contempora-neous nonfinancial corporate debt ratio, as well as the con-temporaneous expected excess portfolio return. The response coefficients of these relationships vary across econ-omies with the size of their bank credit exposures and non-financial corporate debt loads.

The fiscal balance ratio ,

, ,

�FBi tPi tYi tY

depends on a weighted

average of the past short- term nominal market interest rate and the effective long- term nominal market interest rate, as well as the past net government asset ratio, and the contem-poraneous growth rate of nominal output and the primary fiscal balance ratio, according to government dynamic bud-get constraint:

1 1

1ˆ ˆ (1 )

ˆln

ˆ ˆ

ˆ ˆ .,

, ,

,

, 1

,

, 1, ,

, 1 , 1

, ,

, 1 , 1

,

, ,

FB

P Y g

A

P Y

B

Ai

B

Ai

A

P Y

A

P Y

P Y

P Y

PB

P Yi t

i tY

i t

iG

iY

i

iS G

iG i t

S iL G

iG i t

L E i tG

i tY

i t

iG

iY

i

i tY

i t

i tY

i t

i t

i tY

i t

� �

ββ=

++

+ − −

+− −− − − −

ˆ ˆ (1 )ˆ

lnˆ ˆ

ˆ ˆ .,

, 1

,

, 1, ,

, 1 , 1

, ,

, 1 , 1

,

, ,g

A

P Y

B

Ai

B

Ai

A

P Y

A

P Y

P Y

P Y

PB

P YiG

iY

i

iS G

iG i t

S iL G

iG i t

L E i tG

i tY

i t

iG

iY

i

i tY

i t

i tY

i t

i t

i tY

i t

�β+

+ − −

+− −− − − −

In addition, the primary fiscal balance ratio ,

, ,

�PBi tPi tYi tY

de-

pends on the contemporaneous tax rate and the deviation of public domestic demand from output, as well as the terms of trade according to:

�

τ= − −

ˆ ln

ˆ

ˆln

ˆ

ˆ.

,

, ,,

,

,

,

,

PB

P Y

G

Y

G

Y

X

Yi t

i tY

i ti t

i

i

i t

i t

i

i

i tX

i tM

(4.35)

Furthermore, the net government asset ratio ˆ

, 1

, ,

+Ai tPi tYi t

G

Y de-

pends on its past value, as well as the contemporaneous

(4.34)



Furthermore, the output price markup ln ˆ,i t

Yθ , import price markup ln ˆ

,i tMθ , wage markup ln ˆ

,i tLθ , and commodity price

markup ln ˆ,k t

Yθ shocks follow white noise processes:

θ ε ε σ= θ θθln ˆ , ~ iid (0, ),, , ,2

i tY

i t i tY Y

Y (4.52)

θ ε ε σ= θ θθln ˆ , ~ iid (0, ),, , ,2

i tM

i t i tM M

M (4.53)

θ ε ε σ= θ θθln ˆ , ~ iid (0, ),, , ,2

i tL

i t i tL L

L (4.54)

θ ε ε σ= θ θθln ˆ , ~ iid (0, )., , ,2

,k tY

k t k tY Y

Y k (4.55)

Finally, the monetary policy ˆ ,i tiPν , fiscal expenditure ˆ ,i t

Gν , fis-cal revenue ˆ ,i t

Tν , capital requirement ˆ ,i tνκ , and default rate ˆ ,i tνδ shocks follow white noise processes:

ν ε ε σ= ν νν

ˆ , ~ iid (0, ),, , ,2, ,

,i ti

i t i tP i P i P

i P (4.56)


ˆ , ~ iid (0, ),, , ,2

i tG

i t i tG G

G (4.57)


ˆ , ~ iid (0, ),, , ,2

i tT

i t i tT T

T (4.58)

ν ε ε σ=κ ν νν

κ κκˆ , ~ iid (0, ),, , ,

2i t i t i t (4.59)

ν ε ε σ=δ ν νν

δ δδˆ , ~ iid (0, )., , ,

2i t i t i t (4.60)

As an identifying restriction, all innovations are assumed to be independent, which combined with our distributional as-sumptions, implies multivariate normality.

3. ESTIMATIONThe traditional econometric interpretation of an approxi-mate linear state space representation of this New Keynesian DSGE model of the world economy regards it as a represen-tation of the joint probability distribution of the data. We employ a Bayesian maximum likelihood estimation proce-dure that respects this traditional econometric interpretation while conditioning on prior information concerning the generally common values of structural parameters across economies. In addition to mitigating potential model mis-specification and identification problems, exploiting this ad-ditional information may be expected to yield efficiency gains in estimation.

Data Transformations

Estimation of the structural parameters of our New Keynes-ian DSGE model is based on the estimated cyclical compo-nents of a total of 661 endogenous variables observed for 40 economies over the sample period (the first quarter of 1999 through the third quarter of 2014). The advanced and emerging economies under consideration are Argentina, Australia, Austria, Belgium, Brazil, Canada, Chile, China, Colombia, the Czech Republic, Denmark, Finland, France, Germany, Greece, India, Indonesia, Ireland, Israel, Italy, Japan, Korea, Malaysia, Mexico, the Netherlands, New Zealand, Norway, the Philippines, Poland, Portugal, Russia,

price of commodities according to commodity price Phillips curve:

EP E

∑

∑ ∑

ββ

βω ω βω β

φφ

θθ

=+

++

+− −

+

+−

− −

− −−

−

− +=

=

−

=

ln ˆ 1

1ln ˆ

1E ln ˆ (1 )(1 )

(1 )ln

ˆ ˆ

ˆ ˆ

1

11 ln ˆ ln

ˆ ˆ

ˆ1

1ln ˆ ( ) ln ˆ .

, , 1 , 1, ,

, ,1

,

1

*1

,, *, ,

,, 5 , *,

1

P P P wW L

P Y

X

Y

X

XY

P

PL w

k tY

k tY

t k tY k

YkY

kY i

Y i t i t

i tY

i ti

N

kF

i

i

i k

ikF

k

M

i ti i t k t

Y

i tY Y k t

YiY

i i ti

N

EP E

∑

∑ ∑

ββ

βω ω βω β

φφ

θθ

++

++

− −+

+−

− −

− −−

−

− +=

=

−

=

1ln ˆ

1E ln ˆ (1 )(1 )

(1 )ln

ˆ ˆ

ˆ ˆ

1

11 ln ˆ ln

ˆ ˆ

ˆ1

1ln ˆ ( ) ln ˆ .

1 , 1, ,

, ,1

,

1

*1

,, *, ,

,, 5 , *,

1

P wW L

P Y

X

Y

X

XY

P

PL w

k tY

t k tY k

YkY

kY i

Y i t i t

i tY

i ti

N

kF

i

i

i k

ikF

k

M

i ti i t k t

Y

i tY Y k t

YiY

i i ti

NEP E

∑

∑ ∑

ββ

βω ω βω β

φφ

θθ

=+

++

+− −

+

+−

− −

− −−

−

− +=

=

−

=

ln ˆ 1

1ln ˆ

1E ln ˆ (1 )(1 )

(1 )ln

ˆ ˆ

ˆ ˆ

1

11 ln ˆ ln

ˆ ˆ

ˆ1

1ln ˆ ( ) ln ˆ .

, , 1 , 1, ,

, ,1

,

1

*1

,, *, ,

,, 5 , *,

1

P P P wW L

P Y

X

Y

X

XY

P

PL w

k tY

k tY

t k tY k

YkY

kY i

Y i t i t

i tY

i ti

N

kF

i

i

i k

ikF

k

M

i ti i t k t

Y

i tY Y k t

YiY

i i ti

N

EP E

∑

∑ ∑

ββ

ω ω βω β

φθ

θ

++

− −+

− −−

−

=

=

−

=

1E ln ˆ (1 )(1 )

(1 )ln

ˆ ˆ

ˆ ˆ

1 ln ˆ lnˆ ˆ

ˆ1

1ln ˆ ( ) ln ˆ .

1, ,

, ,1

,

1

*1

,, *, ,

,, 5 , *,

1

P wW L

P Y

X

Y

X

XY

P

PL w

t k tY k

YkY

kY i

Y i t i t

i tY

i ti

N

i

i

i k

ikF

k

M

i ti i t k t

Y

i tY Y k t

YiY

i i ti

N

(4.40)

The price of commodities also depends on the contempora-neous, past, and expected future world output weighted av-erage nominal bilateral exchange rate, where the polynomial in the lag operator ( ) 1 1

1 151 = − − Εβ

ββ+ +

−L L Lt . The re-sponse coefficients of this relationship vary across commod-ity markets 1 ≤ k ≤ M*, with k = 1 for energy commodities and k = 2 for nonenergy commodities.

Exogenous Variables

The productivity ln ˆ,i t , labor supply ln ˆ ,i t

Lν , consumption demand ln ˆ ,i t

Cν , investment demand ln ˆ ,i tIν , export demand

ln ˆ ,i tXν , and import demand ln ˆ ,i t

Mν shocks follow stationary first- order autoregressive processes:

i t i t i t i tA A NAA A

Aln ˆ ln ˆ , ~ iid (0, ),, , 1 , ,2ρ ε ε σ= +− (4.41)

ν ρ ν ε ε σ= +νν ν

ν−ln ˆ ln ˆ , ~ iid (0, ),, , 1 , ,2

i tL

i tL

i t i tLL L

L (4.42)


ν−i tC

i tC

i t i tCC C

Cln ˆ ln ˆ , ~ iid (0, ),, , 1 , ,2 (4.43)


ν−ln ˆ ln ˆ , ~ iid (0, ),, , 1 , ,2

i tI

i tI

i t i tII I

I (4.44)


ν−ln ˆ ln ˆ , ~ iid (0, ),, , 1 , ,2

i tX

i tX

i t i tXX X

X (4.45)


ν−ln ˆ ln ˆ , ~ iid (0, )., , 1 , ,2

i tM

i tM

i t i tMM M

M (4.46)

In addition, the credit risk premium ˆ ,i tiSν , duration risk pre-

mium ln ˆ ,i tBν , equity risk premium ln ˆ ,i t

Sν , currency risk pre-mium ln ˆ ,ν ε

i t , and lending rate markup ln ˆ,i t

Cθ shocks follow stationary first- order autoregressive processes:


ν−ˆ ˆ , ~ iid (0, ),, , 1 , ,2

,, ,

,i ti

i ti

i t i tS

i SS i S i S

i S (4.47)


ν−ln ˆ ln ˆ , ~ iid (0, ),, , 1 , ,2

i tB

i tB

i t i tBB B

B (4.48)


ν−ln ˆ ln ˆ , ~ iid (0, ),, , 1 , ,2

i tS

i tS

i t i tSS S

S (4.49)

N Eν ρ ν ε ε σ= +νν ν

ν−ln ˆ ln ˆ , ~ iid (0, ),,E

, 1E

, ,2

EE E

i t i t i t i t (4.50)

θ ρ θ ε ε σ= +θθ θ

θln ˆ ln ˆ , ~ iid (0, )., , , ,2

i tC

i tC

i t i tCC C

C (4.51)


Francis Vitek 71

Poland, Sweden, the United Kingdom, and the United States; by a managed exchange rate regime in Argentina, Brazil, China, Colombia, India, Indonesia, Korea, Malaysia, the Philippines, Russia, South Africa, Switzerland, Thailand, and Turkey; and by a fixed exchange rate regime in Den-mark and Saudi Arabia, consistent with the de facto classifi-cation in IMF 2013. The high- debt contagion economies are Argentina, Brazil, Colombia, Indonesia, Mexico, the Philip-pines, Poland, Russia, South Africa, Thailand, and Turkey, while the low- debt contagion economies are Chile, China, India, and Malaysia. The high- equity contagion economies are Argentina, Brazil, Colombia, India, Indonesia, Mexico, the Philippines, Poland, Russia, South Africa, Thailand, and Turkey, while the low- equity contagion economies are Chile, China, and Malaysia. The quotation currency for transac-tions in the foreign exchange market is issued by the United States. All macroeconomic and financial great ratios are cali-brated to match their observed values in 2012. The same is true of all bilateral trade, bank lending, nonfinancial corpo-rate borrowing, portfolio debt investment, and portfolio eq-uity investment weights, normalized to sum to one across economies.

Parameter estimation results based on the effective sam-ple period (the third quarter of 1999 through the third quar-ter of 2014) are reported in Appendix Table 4.2.1 and Appendix Table 4.2.2. The posterior means of most param-eters are close to their prior means, reflecting the imposition of tight priors to preserve empirically plausible impulse re-sponses. Nevertheless, the data are quite informative regard-ing some of these parameters, as evidenced by substantial updates from prior to posterior, which collectively result in substantial updates to impulse responses.

4. SCENARIO ANALYSISWe illustrate the simulation of banking sector solvency stress test scenarios using the GFM with reference to an example for the United Kingdom, which was an input into IMF 2016. This stress test scenario for the United Kingdom features a balance sheet recession there triggered by monetary normal-ization in the United States, and was the first such applica-tion of the GFM, necessitated by the large cross- border balance sheet exposures and contagion effects involved. The GFM has since been used to simulate stress test scenarios for many other economies, but this example is representative of the general approach.

Scenario Assumptions

Our stress test scenario for the United Kingdom consists of three layers, two external and one domestic (Table 4.1). The external layers are driven by foreign shocks that impact banking sector profitability and capitalization directly by in-creasing funding costs and foreign loan impairments, as well as indirectly through macro- financial spillovers that raise domestic loan impairments. The domestic layer amplifies

Saudi Arabia, South Africa, Spain, Sweden, Switzerland, Thailand, Turkey, the United Kingdom, and the United States. The observed macroeconomic and financial market variables under consideration are the price of output, the price of consumption, the quantity of output, the quantity of private consumption, the quantity of exports, the quan-tity of imports, the nominal policy interest rate, the short- term nominal market interest rate, the nominal bank lending interest rate, the long- term nominal market interest rate, the price of equity, the nominal wage, the unemployment rate, employment, the nominal bilateral exchange rate, the quan-tity of public domestic demand, the fiscal balance ratio, and the prices of nonrenewable energy and nonenergy commodi-ties. For a detailed description of this multivariate panel data set, refer to Appendix 4.1.

We estimate the cyclical components of all of the ob-served endogenous variables under consideration with the generalization of the filter described in Hodrick and Prescott 1997 due to Vitek 2014, which parameterizes the difference order associated with the penalty term determin-ing the smoothness of the trend component. For the price of output, the price of consumption, the quantity of output, the quantity of private consumption, the quantity of exports, the quantity of imports, the price of equity, the nominal wage, employment, the nominal bilateral exchange rate, the quantity of public domestic demand, and the prices of en-ergy and nonenergy commodities, we set the difference or-der to two and the smoothing parameter to 16,000. For the nominal policy interest rate, the short- term nominal market interest rate, the nominal bank lending interest rate, the long- term nominal market interest rate, the unemployment rate, and the fiscal balance ratio, we set the difference order to one and the smoothing parameter to 400.

Parameter Estimates

We estimate the structural parameters of an approximate linear state space representation of our New Keynesian DSGE model by Bayesian maximum likelihood, conditional on prior information concerning their generally common values across economies. Inference on these parameters is based on an asymptotic normal approximation to the poste-rior distribution around its mode, which is calculated by nu-merically maximizing the logarithm of the posterior density kernel with a customized implementation of the differential evolution algorithm due to Storn and Price 1997. We assume a multivariate normal prior distribution, which implies that the mode of the posterior distribution equals its mean. For a detailed discussion of this estimation procedure, refer to Vitek 2014.

The marginal prior distributions of parameters are cen-tered within the range of estimates reported in the existing empirical literature, where available. The conduct of mone-tary policy is represented by a flexible inflation targeting re-gime in Australia, Canada, Chile, the Czech Republic, the euro area, Israel, Japan, Mexico, New Zealand, Norway,



and propagates these adverse impacts. All of these shocks oc-cur relative to the baseline.

The first external layer of our stress test scenario for the United Kingdom features a disorderly accelerated monetary normalization in the United States. In particular, it assumes a 200- basis- point nominal policy interest rate increase in the United States during 2016 and 2017, induced by a strong private domestic demand- driven macroeconomic expansion. Private investment rises four times as much as private con-sumption, driven by investment and consumption demand shocks. This accelerated monetary normalization is accom-panied by a drying up of liquidity in the money market in the United States, reflecting heightened policy uncertainty, represented by a widening of the spread of the short- term nominal market interest rate over the nominal policy inter-est rate by 50 basis points during 2016, driven by interna-tionally correlated credit risk premium shocks. It is also accompanied by an initial steepening of the yield curve in the United States, with the long- term nominal market inter-est rate rising by 200 basis points during 2016, reflecting a rebound of the term premium driven by internationally cor-related duration risk premium shocks that shift investor preferences away from long- term bonds. Furthermore, there is a stock market correction in the United States, with the real equity price falling by 20 percent during 2016, driven by internationally correlated equity risk premium shocks that shift investor preferences away from equities. Finally, the dollar appreciates by 10 percent in real effective terms during 2016, driven by currency risk premium shocks that shift investor preferences toward dollar- denominated finan-cial assets.

The second external layer of our stress test scenario fea-tures financial stress in the “Fragile Four” economies (Brazil,

Indonesia, South Africa, Turkey). Given their high vulnera-bility to monetary normalization in the United States, we assume these four countries experience sudden stops, char-acterized by large domestic demand- driven macroeconomic contractions associated with a tightening of financial condi-tions. In particular, we assume autonomous 4 percent reduc-tions in private investment driven by investment demand shocks, and autonomous 1 percent declines in private con-sumption driven by consumption demand shocks, during 2016 and 2017. Furthermore, we assume 200- basis- point increases in the long- term nominal market interest rate, driven by internationally correlated duration risk premium shocks, and 40 percent reductions in the real equity price driven by internationally correlated equity risk premium shocks, during 2016. In addition, we assume procyclical expenditure- based fiscal consolidation reactions to public debt sustainability concerns, which raise the primary fiscal balance ratio by 2 percentage points during 2016 and 2017. Finally, we assume 10 percent real depreciations of the real, rupiah, rand, and lira against the dollar during 2016, driven by currency risk premium shocks.

The domestic layer of our stress test scenario amplifies the macro- financial impact on the United Kingdom of this mon-etary normalization in the United States. It assumes an au-tonomous private domestic demand- driven macroeconomic contraction in the United Kingdom, featuring a 12 percent reduction in private investment driven by investment de-mand shocks, and a 4 percent decline in private consumption driven by consumption demand shocks, during 2016 and 2017. This autonomous private domestic demand contraction reflects large residential and commercial property market corrections in the United Kingdom, as well as confidence losses by households and firms. The domestic layer of our

TABLE 4.1

Scenario AssumptionsLayer 1: Disorderly Accelerated Monetary Normalization in United States, 2016–2017

Nominal policy interest rate; investment and consumption demand shocks +200 basis pointsMoney market interest rate spread; credit risk-premium shocks +50 basis pointsLong-term nominal market interest rate; duration risk-premium shocks +200 basis pointsReal equity price; equity risk-premium shocks −20 percentReal effective exchange rate; currency risk-premium shocks −10 percent

Layer 2: Financial Stress in Fragile Four, 2016–2017Private investment; investment demand shocks −4 percentPrivate consumption; consumption demand shocks −1 percentLong-term nominal market interest rate; duration risk-premium shocks +200 basis pointsReal equity price; equity risk-premium shocks −40 percentPrimary fiscal balance ratio; fiscal expenditure shocks +2 percentage pointsReal bilateral exchange rate; currency risk-premium shocks +10 percent

Layer 3: Property and Equity Market Corrections in United Kingdom, 2016–2017Private investment; investment demand shocks −12 percentPrivate consumption; consumption demand shocks −4 percentMoney market interest rate spread; credit risk-premium shocks +100 basis pointsLong-term nominal market interest rate; duration risk-premium shocks +100 basis pointsReal equity price; equity risk-premium shocks −40 percent

Source: Author.Note: All scenario assumptions are expressed as deviations from the October 2015 World Economic Outlook baseline (IMF 2015).


Francis Vitek 73

stress test scenario also assumes a drying up of liquidity in the money market in the United Kingdom, reflecting coun-terparty credit risk concerns driven by internationally corre-lated credit risk premium shocks, with the spread of the short- term nominal market interest rate over the nominal policy interest rate widening by 100 basis points during 2016. Furthermore, it assumes a decompression of the term pre-mium in the United Kingdom driven by internationally cor-related duration risk premium shocks, with the long- term nominal market interest rate rising by 100 basis points dur-ing 2016. Finally, it assumes a large stock market correction in the United Kingdom driven by internationally correlated equity risk premium shocks, with the real equity price falling by 40 percent during 2016.

We constrain monetary and fiscal policy responses, as well as bank credit- supply behavior, under this stress test sce-nario. Conventional monetary policy responds endogenously with nominal policy interest rate cuts subject to zero lower bound constraints worldwide. However, we abstract from unconventional monetary policy responses worldwide, and assume that quantitative easing programs remain at their baseline scales in the euro area and Japan. Furthermore, au-tomatic fiscal stabilizers are not allowed to operate in the United Kingdom, given nascent public debt sustainability concerns there. We also abstract from fiscal stimulus mea-sures worldwide. Finally, we constrain bank credit- supply behavior in the United Kingdom through macroprudential policy loosening, roughly equating bank credit growth to nominal output growth. In particular, we abstract from nominal bank lending interest rate increases there in re-sponse to bank capital ratio declines, ensuring that nominal bank lending interest rate adjustments only reflect bank funding costs and credit risk, by reducing regulatory capital requirements.

Simulation Results

Under this stress test scenario, the United Kingdom experi-ences a deep recession exacerbated by an induced contrac-

tion in bank credit supply. Indeed, output falls 7.5 percent below baseline by 2017, while consumption price inflation falls 2.9 percentage points below, and the unemployment rate rises 2.1 percentage points above. Of this output loss, about one third is accounted for by spillovers from the disor-derly accelerated monetary normalization in the United States and the financial stress it triggers in the Fragile Four economies. This deep recession induces a nominal policy in-terest rate cut of 1.4 percentage points below baseline by 2017. But the assumed increase in bank funding costs, in the form of a higher money market interest rate spread, together with the rise in the credit- loss rate, induce a 1.8 percentage point increase in the nominal bank lending interest rate above baseline by 2017, and bank credit falls 7.8 percent be-low. Finally, the fiscal balance ratio falls 1.3 percentage points below baseline by 2017, and the government debt ra-tio rises 11.1 percentage points above, while the current ac-count balance ratio rises 3.5 percentage points above (Figure 4.1 and Figure 4.2).

In the rest of the world, the macroeconomic expansion in the United States is offset by contractions in other advanced economies and vulnerable emerging economies. In particu-lar, while output rises 4.3 percent above baseline in the United States by 2017, it falls 2.9 percent below in other ad-vanced economies, 10.7 percent below in the Fragile Four economies, and 1.3 percent below in other emerging econo-mies. These output losses reflect private domestic demand contractions due to tighter global financial conditions, which generally dominate gains from higher net exports. In aggregate, world output falls 1.4 percent below baseline by 2017, while energy and nonenergy commodity prices fall 20.5 and 14.2 percent below respectively, largely reflecting appreciation of the dollar in nominal effective terms.

5. CONCLUSIONThis chapter presents the GFM, the theoretical structure and empirical properties of which are fully documented in Vitek 2015. It also discusses the application of the GFM to

Source: Author.Note: The gray shade represents countries for which simulation results are not available. There is no country that falls into the 2.0 to 4.0 percent category.

Figure 4.1 Simulated Peak Output Effects

Less than –6.0 –6.0 to –3.0 –3.0 to 0.0 0.0 to 2.0 2.0 to 4.0 More than 4.0Percent



Source: Author.Note: Figure depicts variable paths expressed as output weighted average deviations from baseline. Real effective exchange rate increases represent currency depreciations in real effective terms. ■ = Fragile Four; ■ = other advanced economies; ■ = other emerging economies; ■ = United Kingdom; ■ = United States.

Figure 4.2 Simulation Results

–5.020

15:Q

420

16:Q

420

17:Q

420

18:Q

420

19:Q

420

20:Q

4

5.0

3.0

–1.0

4.0

1.0

–3.0

2.0

–2.0

0.0

–4.0

3.0

–1.0

4.0

1.0

–3.0

2.0

–2.0

0.0

–4.0

Perc

enta

ge P

oint

s

1. Consumption Price Inflation

–15.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

15.0

5.0

0.0

10.0

–5.0

–10.0

Perc

ent

2. Output

–30.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

30.0

10.0

0.0

20.0

–10.0

–20.0

Perc

ent

6. Imports

–10.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

10.0

4.0

0.0

8.0

–2.0

–6.0

2.0

6.0

–4.0

–8.0Perc

enta

ge P

oint

s

7. Policy Interest Rate

–60.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

60.0

0.0

40.0

–20.0

20.0

–40.0

Perc

ent

11. Real Equity Price

–20.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

20.0

0.0

–10.0

10.015.0

5.0

–5.0

–15.0Pe

rcen

t

12. Real Effective Exchange Rate

–10.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

10.0

4.0

0.0

8.0

–2.0

–6.0

2.0

6.0

–4.0

–8.0Perc

enta

ge P

oint

s

8. Short-Term Market Interest Rate

–15.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

15.0

5.0

0.0

10.0

–5.0

–10.0

Perc

ent

3. Consumption

–40.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

40.0

10.00.0

30.0

–10.0–20.0

20.0

–30.0

Perc

ent

4. Investment

–6.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

6.0

2.0

4.0

–2.0

0.0

–4.0

Perc

ent

5. Exports

–5.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

5.0

Perc

enta

ge P

oint

s

9. Bank Lending Interest Rate

3.0

–1.0

1.0

–3.0

2.0

–2.0

0.0

–4.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

4.0

Perc

enta

ge P

oint

s

10. Long-Term Market Interest Rate

3.0

–1.0

1.0

–3.0

2.0

–2.0

0.0

–4.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

4.0

Perc

enta

ge P

oint

s

14. Unemployment Rate

1.5

–0.5

0.5

–1.5

1.0

–1.0

0.0

–2.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

2.0

Perc

enta

ge P

oint

s

15. Fiscal Balance Ratio

4.0

–1.0

2.0

–2.0

1.0

–4.0

3.0

–3.0

0.0

–5.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

5.0

Perc

enta

ge P

oint

s

16. Current Account Balance Ratio

15.0

–5.0

20.0

5.0

–15.0

10.0

–10.0

0.0

–20.0–25.0

2015

:Q4

2016

:Q4

2017

:Q4

2018

:Q4

2019

:Q4

2020

:Q4

25.0

Perc

ent

13. Bank Credit

simulating banking sector solvency stress test scenarios, with reference to an example for the United Kingdom.

The GFM consolidates existing theoretical and empirical knowledge concerning business cycle dynamics in the world economy, provides a framework for a progressive research

strategy, and suggests explanations for its own deficiencies. Future model development work will focus on extending or refining its macro- financial linkages and spillover transmis-sion channels.


Appendix 4.1.Data Description

Estimates are based on quarterly data on a variety of macroeconomic and financial market variables observed for 40 economies over the sample period (the first quarter of 1999 through the third quarter of 2014). The economies under consideration are Argentina, Australia, Austria, Belgium, Brazil, Canada, Chile, China, Colombia, the Czech Republic, Denmark, Finland, France, Germany, Greece, India, Indonesia, Ireland, Israel, Italy, Japan, Korea, Malaysia, Mexico, the Netherlands, New Zea-land, Norway, the Philippines, Poland, Portugal, Russia, Saudi Arabia, South Africa, Spain, Sweden, Switzerland, Thailand, Turkey, the United Kingdom, and the United States. Where available, this data was obtained from the Global Data Source and World Economic Outlook databases compiled by the IMF. Otherwise, it was extracted from the International Financial Statis-tics database produced by the IMF.

The macroeconomic variables under consideration are the price of output, the price of consumption, the quantity of output, the quantity of private consumption, the quantity of exports, the quantity of imports, the nominal wage, the unemployment rate, employment, the quantity of public domestic demand, the fiscal balance ratio, and the prices of nonrenewable energy and nonenergy commodities. The price of output is measured by the seasonally adjusted gross domestic product price deflator, while the price of consumption is proxied by the seasonally adjusted consumer price index. The quantity of output is measured by seasonally adjusted real gross domestic product, while the quantity of private consumption is measured by seasonally ad-justed real private consumption expenditures. The quantity of exports is measured by seasonally adjusted real export revenues, while the quantity of imports is measured by seasonally adjusted real import expenditures. The nominal wage is derived from the quadratically interpolated annual labor income share, while the unemployment rate is measured by the seasonally adjusted share of total unemployment in the total labor force, and employment is measured by quadratically interpolated annual total employment. The quantity of public domestic demand is measured by the sum of quadratically interpolated annual real con-sumption and investment expenditures of the general government, while the fiscal balance is measured by the quadratically interpolated annual overall fiscal balance of the general government. The prices of energy and nonenergy commodities are proxied by broad commodity price indexes denominated in United States dollars.

The financial market variables under consideration are the nominal policy interest rate, the short- term nominal market in-terest rate, the nominal bank lending interest rate, the long- term nominal market interest rate, the price of equity, and the nominal bilateral exchange rate. The nominal policy interest rate is measured by the central bank policy rate, the short- term nominal market interest rate is measured by a reference bank deposit rate, the nominal bank lending interest rate is measured by a reference bank lending rate, and the long- term nominal market interest rate is measured by a long- term government bond yield. In cases where these interest rates are not reported, the closest available substitute is used. The price of equity is proxied by a broad stock price index denominated in domestic currency units, while the nominal bilateral exchange rate is measured by the domestic currency price of one US dollar. All of these financial market variables are expressed as period average values.

Calibration is based on annual data obtained from databases compiled by the IMF where available, and from the Bank for International Settlements (BIS) or the World Bank Group otherwise. Macroeconomic great ratios are derived from the World Economic Outlook and World Development Indicators databases, while financial great ratios are also derived from the Inter-national Financial Statistics and BIS databases. Bilateral trade weights are derived for goods on a cost, insurance, and freight basis from the Direction of Trade Statistics database. Bilateral bank lending and nonfinancial corporate borrowing weights are derived on a consolidated ultimate risk basis from the BIS database. Bilateral portfolio debt and equity investment weights are derived from the Coordinated Portfolio Investment Survey, BIS, and World Development Indicators databases.


Appendix 4.2.Parameter Estimates

APPENDIX TABLE 4.2.1

Parameter Estimation Results (Endogenous variables)

Prior MeanStandard Deviation Posterior Mean Prior Mean

Standard Deviation Posterior Mean

0.9000 0.0900 0.8762 t 0.7500 0.0750 0.7666 0.9756 0.0000 … k 0.7500 0.0000 …χ 2.2500 0.2250 2.0775 d 0.7500 0.0000 …χ B 10.0000 1.0000 10.6072 6.0000 0.6000 6.0322χ C 1.1250 0.1125 1.2115 θY 7.6667 0.0000 …χ G 0.9639 0.0000 … θM 7.6667 0.0000 … 0.0250 0.0000 … θL 7.6667 0.0000 …C 0.0063 0.0000 … θC 161.0000 0.0000 … 0.0500 0.0050 0.0476 ξ0

1.5000 0.1500 1.5381K 0.5000 0.0500 0.5004 ξ1

1.5000 0.1500 1.4182C 0.5000 0.0500 0.5101 ξ0

Y 0.1250 0.0125 0.1257Y 0.5000 0.0500 0.4914 ξ1

Y 0.1250 0.0125 0.1244 M 0.5000 0.0500 0.5223 ξ1

Q 0.0313 0.0031 0.0305 L 0.5000 0.0500 0.5278 ξ2 1.2500 0.0000 …i 0.0800 0.0080 0.0834 i 0.0000 0.0000 …R 0.1000 0.0000 … G 0.0025 0.0000 …M 0.2500 0.0250 0.2590 0.0250 0.0000 …ωY 0.8750 0.0875 0.8717 ,B 0.0625 0.0000 …ωM 0.8750 0.0875 0.8803 ,i 0.1250 0.0000 …ωL 0.8750 0.0875 0.8558 ,B 0.0156 0.0000 …ωC 0.3333 0.0333 0.3334 ,i 0.1250 0.0000 …ω1

Y 0.3333 0.0333 0.3227 0M 0.5409 0.0541 0.5044

ω2Y 0.3333 0.0333 0.3534 1

M 1.0817 0.1082 0.9922 0.8000 0.0800 0.7975 2

M 1.6226 0.1623 1.6199A 0.1000 0.0100 0.0977 0

B 0.5409 0.0541 0.5394C 0.4500 0.0450 0.4359 1

B 1.0817 0.1082 1.04101

F 0.9000 0.0900 0.9255 2B 1.6226 0.1623 1.5839

2F 0.8000 0.0800 0.8225 0

S 0.6649 0.0665 0.6405ψ M 1.5000 0.1500 1.5422 1

S 1.3297 0.1330 1.3220 i 0.7500 0.0750 0.7755 2

S 1.9946 0.1995 2.0544

G 0.7500 0.0750 0.7727

Source: Author.Note: All priors are normally distributed, while all posteriors are asymptotically normally distributed. All auxiliary parameters have degenerate priors with mean zero.




Parameter Estimation Results (Exogenous variables)

Prior MeanStandard Deviation Posterior Mean Prior Mean

Standard Deviation Posterior Mean

A 0.7500 0.0750 0.7715 2vX 6.19 × 10+0 6.19 × 10−1 6.30 × 10+0

vc 0.5000 0.0500 0.5231 2vM 8.36 × 10+0 8.36 × 10−1 8.56 × 10+0

vl 0.5000 0.0500 0.5270 2vi,P 2.89 × 10−1 2.89 × 10−2 2.91 × 10−1

vx 0.7500 0.0750 0.7134 2vi,S 3.26 × 10−2 3.26 × 10−3 3.20 × 10−2

vM 0.7500 0.0750 0.7536 2vB 2.61 × 10−1 2.61 × 10−2 2.64 × 10−1

vi,S 0.7500 0.0750 0.7449 2vS 4.03 × 10+0 4.03 × 10−1 3.96 × 10+0

vB 0.7500 0.0750 0.8017 2C 1.15 × 10+4 1.15 × 10+3 1.17 × 10+4

vS 0.7500 0.0750 0.7658 2L 3.47 × 10+5 3.47 × 10+4 3.50 × 10+5

θC 0.5000 0.0500 0.4945 2vL 2.14 × 10+1 2.14 × 10+0 2.18 × 10+1

vL 0.7500 0.0750 0.7437 2v 6.02 × 10−2 6.02 × 10−3 5.96 × 10−2

v 0.7500 0.0750 0.7347 2vG 8.20 × 10−2 8.20 × 10−3 7.73 × 10−2

2θY 7.31 × 10+4 7.31 × 10+3 7.29 × 10+4 2

vT 4.33 × 10−1 4.33 × 10−2 4.44 × 10−1

2θM 1.30 × 10+5 1.30 × 10+4 1.35 × 10+5 2

v 1.57 × 10−1 1.57 × 10−2 1.53 × 10−1

2A 4.59 × 10−1 4.59 × 10−2 4.59 × 10−1 2

v 1.07 × 10−3 1.07 × 10−4 1.09 × 10−3

2vC 1.07 × 10+1 1.07 × 10+0 9.79 × 10+0 2

Y,k 7.52 × 10+3 7.52 × 10+2 7.83 × 10+3

2vI 1.77 × 10+0 1.77 × 10−1 1.76 × 10+0

Source: Author.Note: All priors are normally distributed, while all posteriors are asymptotically normally distributed. All auxiliary parameters have degenerate priors with mean zero.


Francis Vitek 79

Kiyotaki, Nobuhiro, and John Moore. 1997. “Credit Cycles.” Jour-nal of Political Economy 105: 211–248.

Monacelli, Tommaso. 2005. “Monetary Policy in a Low Pass- Through Environment.” Journal of Money, Credit, and Banking 37: 1047–1066.

Smets, Frank, and Raf Wouters. 2003. “An Estimated Dynamic Stochastic General Equilibrium Model of the Euro Area.” Jour-nal of the European Economic Association 1: 1123–75.

Storn, Rainer, and Kenneth V. Price. 1997. “Differential Evolution— A Simple and Efficient Heuristic for Global Opti-mization over Continuous Spaces.” Journal of Global Optimiza-tion 11: 341–359.

Tobin, James. 1969. “A General Equilibrium Approach to Mone-tary Theory.” Journal of Money, Credit, and Banking 1: 15–29.

Vitek, Francis. 2014. “Policy and Spillover Analysis in the World Economy: A Panel Dynamic Stochastic General Equilibrium Approach.” IMF Working Paper 14/200, International Mone-tary Fund, Washington, DC. https://www.imf.org/en/Publica-tions/WP/Issues/2016/12/31/ Policy- and- Spillover- Analysis - in- the- World- Economy- A-Panel-Dynamic-Stochastic -General-42433.

———. 2015. “ Macro- Financial Analysis in the World Economy: A Panel Dynamic Stochastic General Equilibrium Approach.” IMF Working Paper 15/227, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ Macrofinancial- Analysis- in- the- World - E c onomy- A - Pa ne l -D y n a m ic - S to c h a s t i c - G e ne r a l -Equilibrium-43369.

REFERENCESChristiano, Lawrence, Martin Eichenbaum, and Charles L. Evans.

2005. “Nominal Rigidities and the Dynamic Effects of a Shock to Monetary Policy.” Journal of Political Economy 113: 1–45.

Galí, Jordi. 2011. “The Return of the Wage Phillips Curve.” Jour-nal of the European Economic Association 9: 436–461.

Gerali, Andrea, Stefano Neri, Luca Sessa, and Federico M. Signo-retti. 2010. “Credit and Banking in a DSGE Model of the Euro Area.” Journal of Money, Credit, and Banking 42: 107–141.

Hodrick, Robert J., and Edward C. Prescott. 1997. “ Post- War US Business Cycles: A Descriptive Empirical Investigation.” Jour-nal of Money, Credit, and Banking 29: 1–16.

Hülsewig, Oliver, Eric Mayer, and Timo Wollmershäuser. 2009. “Bank Behavior, Incomplete Interest Rate Pass- through, and the Cost Channel of Monetary Policy Transmission.” Economic Modelling 26: 1310–1327.

International Monetary Fund (IMF). 2013. Annual Report on Ex-change Arrangements and Exchange Restrictions. IMF: Washing-ton, DC. https://www.imf.org/en/Publications/ Annual- Report - on- Exchange- Arrangements- and- Exchange- Restrictions / I s s u e s /2 016 /12 /31/ A n nu a l - R e p or t- on - E xc h a n g e -Arrangements-and-Exchange-Restrictions-2013-40681.

———. 2015. World Economic Outlook— Adjusting to Lower Com-modity Prices. Washington, DC, October.

———. 2016. “United Kingdom: Financial System Stability As-sessment.” IMF Country Report 16/167, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / United- Kingdom- Financial- Sector- Assessment- Program -Financial-System-Stability-Assessment-43978.


https://www.imf.org/en/Publica-tions/WP/Issues/2016/12/31/




https://www.imf.org/en/Publications/











CHAPTER 5

G20 Data Gaps Initiative II: Meeting the Policy Challenge

ROBERT HEATH • EVRIM BESE GOKSU

The G20 Data Gaps Initiative (DGI), which aimed at addressing the information needs that were revealed by the 2007–08 global financial crisis, concluded its first phase and started a second phase (DGI-2) with the endorsement of G20 Finance Ministers and Central Bank Governors in

September 2015. The DGI-2 recommendations maintain the continuity of DGI-1 but, reflecting the evolving policy needs, focus more on datasets that support the monitoring of risks in the financial sector and the analysis of the interlinkages across the economic and financial systems. This chapter presents the DGI as an overarching initiative, bringing together various statistical frameworks for a complete picture of the economic and financial system to support the work of policymakers.

2. RESPONDING TO POLICY NEEDS: THE DGI

How the Evolution of Economic Thinking over Time Affected Statistics

Recognizing the need to strengthen economic and financial data following a crisis is not new. As the economic and fi-nancial systems evolve as a consequence of market develop-ments and financial innovation, information needs change. Looking back in history, crisis events have always acted as triggers to question the nature, quality, and availability of data needed for policymaking.

The Great Depression of the 1930s is a good example of fundamental advances in economic statistics. As policymak-ers began to more actively manage the economy and particu-larly aggregate demand, the intellectual and policy focus came to be concentrated on demand and supply factors in the economy, and on transactions rather than stocks. As a result, the System of National Accounts (SNA), which still re-mains the overarching framework of macroeconomic statis-tics, was developed in the late 1940s (UN 1953), and the first IMF Balance of Payments Manual (IMF 1948) was pub-lished around the same time.

The capital liberalization trend which started in the 1980s brought new opportunities for investment but also new risks

1. INTRODUCTIONA widely accepted old lesson is that “Good data and good analysis are the lifeblood of effective surveillance and policy responses at both national and international levels” (FSB and IMF 2009). Indeed, reliable, comprehensive, and timely information is essential to assess the risks and vulnerabilities facing economies, as policymaking relies on a correct assess-ment of such risks and vulnerabilities.

In 2007–08 the problems in the financial systems of a number of advanced economies, including the United States, spilled across borders to affect the rest of the world. As the financial sector was at the center of the crisis, the G20 economies supported a number of actions for the re-form of the financial sector regulatory framework. Even though a lack of data was not the main reason for the crisis, it would have been possible to detect risk build-ups had the right data been available at the right time. Further, better data might also have enhanced the robustness and credibil-ity of stress test analysis conducted by the authorities around the crisis, helping to support market confidence. To this end, the identification and addressing of information gaps were among the action items for the reform of the fi-nancial sector leading to the G20 Data Gaps Initiative (DGI). This chapter explains how the DGI is meeting pol-icy needs.

This chapter is based on IMF Working Paper 16/43 (Heath and Bese Goksu 2016).


G20 Data Gaps Initiative II: Meeting the Policy Challenge82

economies, and the close cooperation among relevant par-ties. The InterAgency Group on Economic and Financial Statistics (IAG)1 has been acting as the global facilitator and coordinator of the exercise, liaising with other groups and initiatives. The IMF staff has been monitoring the imple-mentation of the DGI recommendations by G20 economies on an annual basis and reporting, together with the FSB Secretariat, the progress made to the G20 FMCBG. Six progress reports were provided to the G20 FMCBG in DGI-1, with the third progress report under DGI-2 pro-vided in September 2018 (http://www.imf.org/external/ns /cs.aspx?id=290).

In 2009, many of the recommendations were written as aspirations, as the implications of the crisis for regulatory and financial policy going forward were unclear. They were drafted following extensive consultations with compilers and users, including a users’ conference in July 2009,2 and structured around four themes: build-up of risk in the finan-cial sector, cross-border financial vulnerabilities, vulnerabil-ity of domestic economies to shocks, and communication of official statistics.

As time progressed, the implications of the crisis for regu-latory and macroprudential policy, and hence the data needs, have become more clearly established. Reflecting this, the DGI-2 recommendations focus on datasets that support the stability of the financial system both domestically and inter-nationally. Nonetheless, as the 20 recommendations in DGI-1 have stood the test of time, DGI-2 represents an evo-lution and not a rethinking of the DGI project. DGI-2 aims to strengthen and consolidate the progress made in DGI-1, achieve the potential for data provision embodied in the ini-tiative, and promote high-quality statistics for policy use (see Figure 5.1). The DGI-2 recommendations are set out in Appendix 5.1.

The intention of this chapter is to demonstrate how DGI-2 is integral to meeting the emerging policy needs, both regulatory and macro-financial. To this end, the DGI-2 recommendations are more specific than those of DGI-1, with some identified as global priorities (see Figure 5.1) based on consultations with users and compilers in 2015. G20 economies have committed to action plans that take national circumstances into account, but are based on the targets set for each recommendation. The objective is to ad-vance the statistical agenda agreed to by the G20 economies and endorsed by the G20 FMCBG at the global level. This agenda is designed to help make national and international financial systems more stable in a world of increased finan-cial interconnectedness. An earlier working paper that set

and vulnerabilities, domestically and across borders, leading to a growing policy focus on financial stability (Heath 2015). These developments have necessitated a rethinking of mac-roprudential and monetary policies (IMF 2015g) and also the related statistical frameworks.

When the Mexican crisis occurred in 1994–95, interna-tional capital flows and a lack of relevant information were central. The IMF responded with the establishment of two key standards for the dissemination of a core set of economic and financial data: the Special Data Dissemination Standard (SDDS) and the General Data Dissemination System (GDDS). The SDDS was intended for countries with access to international capital markets while the GDDS focused on countries that needed to develop their statistical systems.

In 1997–98 the Asian crisis revealed the need for better information on reserve and reserve-related activities, as for-ward sales of foreign currency contracts by the Bank of Thai-land were seen as having masked the true pressure on international reserves. As a result, a reserves template was de-veloped and the SDDS was strengthened by the addition of requirements on reserves and foreign currency liquidity data.

Due to the global imbalances and the associated discrep-ancies in income flows at the global level, the first IMF Co-ordinated Portfolio Investment Survey (CPIS) was launched at end 1997 to improve statistics of holdings of portfolio in-vestment assets in the form of equity and long-term and short-term debt securities. The strengthening of Bank for International Settlements’s (BIS) International Banking Sta-tistics (IBS) and the increasing adoption by countries of the IMF’s Balance of Payments Manual has been prevalent throughout the past two decades.

The Global Financial Crisis of 2007–08 and the DGI

The financial crisis, which started in 2007 with problems in the US subprime market, spread to the rest of the world and became the most severe global crisis since the Great Depres-sion. One difference between the global financial crisis and earlier postwar crises was that the crisis struck at the heart of the global financial system and spread throughout the global economy. This required global efforts for recovery. As one element of the global response, in October 2009 the G20 Finance Ministers and Central Bank Governors (FM-CBG) endorsed a DGI led by the Financial Stability Board (FSB) Secretariat and the IMF staff. The DGI was launched as an overarching initiative of 20 recommendations to ad-dress information gaps revealed by the global financial crisis.

Since its launch, considerable progress has been made to-ward closing those gaps (FSB and IMF 2015). Given this progress, in September 2015, at its sixth year, the G20 FM-CBG closed the first phase (DGI-1) and opened a second act of the DGI (DGI-2).

The success of the DGI is mainly attributable to a strong policy support, a common sense of ownership by the G20

1 The IAG members are the BIS, European Central Bank, Eurostat, IMF (chair), Organisation for Economic Co-operation and Development, United Nations, and World Bank. The FSB is invited to participate on topics in which it has a direct involvement.

2 Papers from the IMF and Financial Stability Board’s 2009 Users’ Con-ference on the Financial Crisis and Information Gaps are available at http://www.imf.org/external/np/seminars/eng/2009/usersconf/index .htm.


http://www.imf.org/external/ns/cs.aspx?id=290

http://www.imf.org/external/ns/cs.aspx?id=290

http://www.imf.org/external/np/seminars/eng/2009/usersconf/index.htm

http://www.imf.org/external/np/seminars/eng/2009/usersconf/index.htm

Robert Heath and Evrim Bese Goksu 83

was established in April 2009 as the successor to the Finan-cial Stability Forum and started working as the central locus of coordination to take forward the financial reform pro-gram as developed by the relevant bodies. The obligations of members of the FSB included agreeing to undergo periodic peer reviews, using among other inputs IMF/World Bank Financial Sector Assessment Program (FSAP) reports. The G20 leaders noted the importance of global efforts in imple-menting the global regulatory reform so as to protect against adverse cross-border, regional, and global developments af-fecting international financial stability.

The components of the G20 regulatory reform agenda complement each other with an ultimate goal of strengthen-ing the international financial system. The DGI has been an important element of this agenda as the regulatory reform agenda items mostly require better data. The collection of data on Global Systemically Important Banks’ (G-SIBs) ex-posures and funding dependencies is among the steps to-wards addressing the “too-big-to-fail” issue by reducing the probability and impact of G-SIBs’ failing. The FSB work on developing standards and processes for global data collection

out the analytical justification for the DGI-1 recommenda-tions remains relevant for DGI-2 (Heath 2013).

Evolution of Policy Needs Following the Global Financial Crisis

Regulatory Reform Agenda

Following the global financial crisis, in 2008, the G20 lead-ers, at their meeting in Washington,3 committed to imple-menting a fundamental reform of the global financial system to strengthen financial markets and regulatory regimes so as to avoid future crises.4 As part of the reform agenda, the FSB

DGI-I Recommendations

Build-up of risk in the financial sectorI.1: Mandate

Cross-border financial linkages

Vulnerability of domestic economies to shocks

Communication of official statistics

I.2: Financial Soundness IndicatorsI.3: Tail riskI.4: Aggregate leverage and maturitymismatchesI.5: Credit default swapsI.6: Structured productsI.7: Securities data

I.8 and I.9: Data for global systemically important financial institutionsI.10 and I.11: Coordinated Portfolio Investment Survey and international banking statistics participation and enhancementI.12: International investment positionI.13 and I.14: Financial and nonfinancial corporations’ cross-border exposures

I.15: Sectoral accountsI.16: Distributional informationI.17: Government Finance StatisticsI.18: Public sector debtI.19: Real estate prices

I.20: Principal Global Indicators

DGI-II Recommendations

Monitoring risks in the financial sectorII.1: Mandate

Vulnerabilities, interconnections, and spillovers

Communication of official statistics

II.2: Financial Soundness Indicators*II.3: Financial Soundness Indicators concentration and distribution measuresII.4: Data for global systemically important financial institutionsII.5: Shadow bankingII.6: DerivativesII.7: Securities statistics*

II.8: Sectoral accounts*II.9: Household distributional informationII.10: International investment positionII.11: International banking statistics*II.12: Coordinated Portfolio Investment Survey*II.13: Coordinated Direct Investment SurveyII.14: Cross-border exposures of nonbank corporationsII.15: Government Finance Statistics*II.16: Public Sector Debt Database*II.17: Residential property pricesII.18: Commercial property prices

II.19: International data cooperation and communicationII.20: Promotion of data sharing

Recommendations that are completed based on the targets that were introduced in 2014.Recommendations where significant progress was made and are close to completionpending participation by all G20.Recommendations where progress was slow.

Source: Author.*Indicates priority areas identified by the G20 economies and international agencies in 2015.

Figure 5.1 Data Gaps Intitiative Recommendations

3 “Declaration Summit on Financial Markets and the World Economy, November 15, 2008.” http://www.un.org/ga/president/63/commission /declarationG20.pdf.

4 The G20 leaders continue to reaffirm the importance of this commit-ment. For instance, at the Antalya summit in November 2015, the lead-ers stated that “Going forward, we are committed to full and consistent implementation of the global financial regulatory framework in line with the agreed timelines….”


http://www.un.org/ga/president/63/commission/declarationG20.pdf

http://www.un.org/ga/president/63/commission/declarationG20.pdf


tion Plan for Strengthening Surveillance following the 2014 TSR (IMF 2014c) underlined that the IMF will revive and adapt the balance sheet approach to facilitate a more in-depth analysis of the impact of shocks and their transmis-sion across sectors, and possibly initiate the global flow of funds to better reflect global interconnections (Box 5.1). This work requires data from the DGI, as it will help support the IMF’s macro-financial work, including in the key exer-cises and reports (that is, early warning exercise, FSAP, and GFSR).

The DGI Project

The DGI project has allowed for a broad range of users’ needs to be incorporated into the development of economic and financial statistics. Wide user consultation took place as part of the DGI work process in 2015,7 including through the Second IMF Statistical Forum, which constitutes an an-nual global space where data users, data providers, and poli-cymakers can come together to discuss emerging needs for statistical information to inform policymaking.8

The consultations with users indicated the need for en-suring completeness and comprehensiveness of data that support analysis of interconnections among economies. The importance of the balance sheet approach for understanding sectoral interconnections within the domestic economy was emphasized. Data that are key to assessing fiscal sustainabil-ity were agreed to be essential but challenges in implementa-tion were also pointed out. As a result, the following key focus areas common for G20 economies were identified:

• Disseminating consistent and comparable Financial Soundness Indicators

• Ensuring regular collection of IBS and the CPIS• Providing consistent securities statistics

and aggregation on securities financing transactions aims to improve transparency in securitization toward the main goal of reducing risks related to the shadow banking system. Over-the-counter (OTC) derivatives markets including credit default swaps (CDS) were brought under greater scru-tiny toward the main goal of making derivatives markets safer following the global crisis. The DGI supported this goal by improving information in CDS markets. A number of other G20 initiatives have strong links with the DGI proj-ect, including the FSB work on strengthening the oversight and regulation of the shadow banking system and on the work on global legal entity identifiers5 that contribute to the robustness of the data frameworks with a more micro focus. The changing global regulatory reforms—particularly the implementation of Basel III—was also taken into consider-ation in the development of the DGI.

Surveillance Agenda

The importance of closing the data gaps hampering the sur-veillance of financial systems was also highlighted as part of the IMF’s 2014 Triennial Surveillance Review (TSR).6 The 2014 TSR emphasized that due to growing interconnected-ness across borders, financial market shocks will continue to have significant spillovers via both capital flows and shifts in risk positions. Also, new dimensions to interconnectedness will continue to emerge, such as through the potential short-term adverse spillovers generated by the financial regulatory reforms. To this end, the TSR recommended improving in-formation on balance sheets and enriching flow-of-funds data.

The IMF has overhauled its surveillance to make it more risk based. To this end, the IMF Managing Director’s Ac-

Box 5.1. Global Flow of Funds

Through the use of internationally agreed-upon statistical standards, data on cross-border financial exposures (International Banking Sta-tistics, Coordinated Portfolio Investment Survey, and Coordinated Direct Investment Survey) can be linked with domestic sectoral ac-counts data to build up a comprehensive picture of financial interconnections domestically and across borders, with a link back to the real economy through the sectoral accounts. This work is known as the “Global Flow of Funds” (GFF) (Errico and others 2014). The GFF project is mainly aimed at constructing a matrix that identifies interlinkages among domestic sectors and with counterpart countries (and possibly counterpart country sectors) to build up a picture of bilateral financial exposures and support analysis of potential sources of contagion.

The concept of the GFF was first outlined in the Second Progress Report on the G20 Data Gaps Initiative and initiated in 2013 as part of a broader IMF initiative aimed at strengthening the analysis of interconnectedness across borders, global liquidity flows, and global financial interdependencies. In the longer term, the GFF matrix is intended to support regular monitoring of bilateral cross-border financial posi-tions through a framework that highlight risks to national and international financial stability. The IMF staff is working toward developing a GFF matrix, starting with the largest global economies.

5 A global legal entity identifiers system would uniquely identify parties to financial transactions.

6 The papers contributing to the review, including the overview paper, are available at http://www.imf.org/external/np/spr/triennial/2014/. The TSR involved wide consultation among IMF member countries, aca-demia, and the private sector.

7 Four regional conferences were held, as well as meetings with private sector participants.

8 The Second IMF Statistical Forum was held November 18–19, 2014, at IMF Headquarters in Washington, DC. Proceedings are available at http://www.imf.org/external/np/seminars/eng/2014/statsforum/.


http://www.imf.org/external/np/spr/triennial/2014/

http://www.imf.org/external/np/seminars/eng/2014/statsforum/


How Does the DGI Help Monitor Risks in the Financial Sector?

Assessing the Soundness of the Banking System

It has been recognized for some time that microanalysis of financial institutions needs to be complemented with a macro focus. To this end, the IMF’s financial soundness indicators (FSIs) were developed in the early 2000s and, while back-ward looking, are an important component of a macropru-dential framework for monitoring and assessing the health and soundness of the overall financial sector (Navajas and Thegeya 2013; and http://www.imf.org/external/np/sta/fsi /eng/fsi.htm ). To date, the main focus of the FSIs has been on the banking sector, with additional indicators on banks’ customers as well as on the markets that they operate in.

From the consultations with compilers and users in 2015, it is clear that FSIs are also increasingly being used by na-tional authorities to establish national benchmarks, perform cross-country analyses, and construct early warning indica-tors. Such analyses are feeding into financial stability reports to inform policymaking. Further, the IMF includes FSI data in individual economy Article IV consultation reports and in the statistical annex of the GFSR. This policy-related focus of FSIs, both at the national and international level, helped encourage a significant increase in country cov-erage of FSIs reported to the IMF during DGI-1. At the end

• Improving the availability of sectoral accounts data• Disseminating timely and comparable general gov-

ernment operations and debt dataThe Second IMF Statistical Forum, under the main theme

of “Statistics for Policy Making—Identifying Macroeconomic and Financial Vulnerabilities,” emphasized the importance of data quality and comparability; the need for monitoring inter-connections and so the importance of sectoral accounts, bal-ance sheets, and international investment position (IIP) data; and the need for better information on nonfinancial corpora-tions, households, and real estate markets.

Consequent to these developments DGI-2 emerged with a focus on: (1) monitoring risk in the financial sector, and (2) vulnerabilities, interconnections, and spillovers. As illus-trated in Figure 5.2, the recommendations in DGI-2 can mostly be clustered under these two broad headings. Further, the recommendations are presented as a coherent package that, in their implementation, create positive externalities for both compilation and analysis. The “vulnerabilities, intercon-nections, and spillovers” category is based on the overarching national accounts system, while the recommendations in the “monitoring risk in the financial system” section cover finan-cial institutions and financial markets, with shadow banking straddling both institutions and markets, as explained in the next section. All recommendations fit together to provide an overall picture of the economy and the financial sector.

I. Monitoring risk in the financial sector

Financial instruments and markets

II. Vulnerabilities, interconnections, and spillovers

Sectoral accounts and balance sheets

Derivatives(R. II.6)

Securities(R. II.7)

FSIs(R. II.2)

CDMs(R. II.3)

ShadowBanking(R. II.5)

Data forG-SIFIs(R. II.4)

Financial institutions

Nonfinancialassets

RPPI(R. II.17)

CPPI(R. II.18)

IIP(R. II.10)

IBS(R. II.11)

CPIS(R. II.12)

CDIS(R. II.13)

GFS(R. II.15)

PSDS(R. II.16)

Nonfinancialcorporations

Householdsector

Fiscaldata

Deposittakers

Sectoralaccounts(R. II.8)


Cross-borderexposures(R. II.14)


Householdsdistributional

data(R. II.9)

Cross-borderexposures(R. II.14)

Externalsector

Nonbank financialcorporations

Source: Author.Note: CDIS = Coordinated Direct Investment Survey; CDMs = concentration and distribution measures; CPIS = Coordinated Portfolio Investment Survey; CPPI = Commercial Property Price Indices; DGI = Data Gaps Initiatives; GFS = Government Finance Statistics; G-SIFIs = global systemic financial institutions; FSIs = Financial Soundness Indicators; IBS = International Banking Statistics; IIP = international investment position; PSDS = Public Sector Debt Statistics; R = DGI Recommendation; RPPI = Residential Property Price Indices.

Figure 5.2 Linkages within the DGI-2 Recommendations


http://www.imf.org/external/np/sta/fsi/eng/fsi.htm



toring financial sector vulnerabilities. In DGI-2, the IMF is discussing regular collection of CDM data (Appendix 5.1, Recommendation II.3).

Regarding the banking sector, the BIS conducted concep-tual work focusing on system-level measures of maturity mis-matches (funding gaps) on banks’ international balance sheets, based on BIS IBS (Fender and McGuire 2010). The BIS’s work pointed out that analysis of system-wide bank funding risks and the transmission of shocks across countries require geographically disaggregated data on banks’ balance sheets to capture funding patterns that are location specific, and facilitate targeted assessments of vulnerabilities showing up in the aggregate data. This work helped inform the en-hancements to the BIS IBS that were adopted during 2012–15, thereby improving the usefulness of this dataset for the construction of maturity mismatch and leverage measures. The enhancements to the IBS included improved information on the counterparty, residual maturity, and currency break-down of banks’ international positions, with improvements made both to the residency-based and consolidated-based (us-ing nationality-based supervisory concepts) statistics. DGI-2 maintains an emphasis on improving the IBS reporting of G20 economies (Appendix 5.1, Recommendation II.11).

Monitoring the Shadow Banks

A shadow banking system is defined by the FSB as a “credit intermediation involving entities and activities outside the regular banking system.” Such institutions could provide al-ternative sources of funding for market participants in com-plement to traditional banking but could also carry bank-like risks. Those risks could easily spread through the rest of the system due to complex relationships among these institu-tions and banks, and hence, they need to be monitored.

Typically, these institutions are highly leveraged and heavily reliant on short-term funding while investing in long-term illiquid assets and hence are exposed to liquidity and maturity risks. During the crisis when such risks materi-alized, the entire financial system suffered the consequences, thus emphasizing the importance of monitoring such risks.

At the 2011 Summit Meeting in Cannes, the G20 leaders asked the FSB to address the financial stability concerns as-sociated with shadow banking. The FSB strategy has two el-ements (see FSB 2013):

• First, the FSB initiated an annual global shadow banking monitoring exercise.

• Second, the FSB is working to develop policies to strengthen oversight and regulation of the shadow banking system.11

The FSB’s 2015 annual report covered 26 jurisdictions which, as of 2014 (FSB 2015b) constituted 80 percent of global GDP and 90 percent of global financial system assets and is based on balance sheet data of national financial

of 2018 almost 140 economies reported FSI data to the IMF, including all G20 economies—an increase from 45 econo-mies in 2009.

However, to maintain the usefulness of FSIs as a tool for financial stability assessment, the list of indicators was up-dated in 2013 to reflect the changes in the financial environ-ment, notably the increased prominence of nonbank financial institutions, and the global regulatory reforms, particularly the implementation of Basel III (see IMF 2013a). The latter revised the definitions of capital and introduced new mea-sures of leverage, liquidity, and funding—all of which are reflected in the updated list of FSIs.

The updated list of FSIs for nonbank financial corpora-tions aims to contribute to the analysis and assessment of the potential impacts of the shadow banking sector on the sta-bility of the financial system. Whereas the previous list only looked at the subsector as a whole, which is comprised of a very heterogeneous set of institutions, the new list includes separate FSIs for money market funds, insurance corpora-tions and pension funds, and other nonbank financial insti-tutions. New FSIs were also introduced for nonfinancial corporations and households (see http://www.imf.org/external /np/sta/fsi/eng/fsi.htm). The DGI-2 places greater emphasis on increasing the frequency and coverage of FSI reporting, particularly for nonbank financial institutions (Appendix 5.1, Recommendation II.2).

The crisis also highlighted the need for taking tail risks into account as a complement to the overall assessment of the financial sector risks through aggregate measures. To this end, to capture the system-wide disturbances that could be caused by the institutions that are at the tail of the distri-butions, aggregate FSI measures were enhanced by a pilot study on concentration and distribution measures (CDMs). It was considered that expanding FSIs for the financial sys-tem with CDMs would allow policymakers and the IMF staff to better capture the performance of the financial sector with greater granularity and in a forward-looking manner.

The pilot project was completed in 2015, with the partici-pation of 35 diverse countries (see Crowley and others 2016). CDMs were compiled for six FSIs of deposit takers: regula-tory Tier 1 capital to risk-weighted assets, nonperforming loans to total gross loans, return on assets, return on equity, liquid assets to short-term liabilities, and capital to total as-sets.9 The data provided important information that was not revealed by averages. For instance, the distributions of mini-mum values of CDMs,10 which represent the institutions with the most severe risks for any variable, showed substan-tial variation across countries and over time within coun-tries. The pilot project indicated that regular reporting of CDMs may be feasible and could be a useful tool for moni-

9 The CDMs included the following indicators: (1) minimum, maxi-mum, and mean; (2) weighted standard deviations and skewnesses; and (3) quartiles, and the asset share of the bottom quartile.

10 Maximum values in the case of the nonperforming loan FSIs.11 The progress on the FSB work on shadow banking is set out in FSB

2015c.





The data are stored at the International Data Hub estab-lished at the BIS and currently are shared among the data-providing national authorities.12 This process has reinforced the exchange of information and coordination among na-tional supervisory authorities. However, given the granular-ity of the dataset, it brings along confidentiality issues that need to be addressed in the longer term to make better use of this critical information.

The objective of the templates is to provide authorities with a clearer view of global financial networks and assist them in their supervisory and macroprudential responsibili-ties (FSB 2014). Data on G-SIBs supports the IMF’s work in safeguarding international financial stability including through effective multilateral and bilateral surveillance and the encouragement of coherent policy responses across mem-ber countries (IMF 2014c). The data permits bank-level information to be used in conjunction with measures of worldwide exposures, substantially improving the ability to detect vulnerabilities that could originate from common ex-posures and concentrated funding positions so deepening the understanding of the potential source of spillovers. G-SIBs data also improves the tracking of banks’ cross-currency funding and maturity transformation activities. In addition, the data helps to improve understanding of financial innova-tion, market complexity, and emerging sources of potential systemic risks.

Going beyond the banking industry, the FSB and Interna-tional Association of Insurance Supervisors have also identi-fied insurance companies of global systemic importance, based on a methodology developed by the International Association of Insurance Supervisors (IAIS 2013). The assess-ment methodology relates to the methodology developed by the Basel Committee on Banking Supervision for G-SIBs, but also takes into account the specific nature of the insurance sector. In particular, insurance groups that engage in nontra-ditional or noninsurance activities can be vulnerable to li-quidity and market price risks amplifying or contributing to systemic risk (IAIS 2011). Therefore, such nontraditional ac-tivities are included as an indicator in the assessment meth-odology. Following on from these regulatory developments, in DGI-2 the possibility of developing a common data tem-plate for global systemically important nonbank financial in-stitutions, starting with insurance companies, is included in DGI-2 (Appendix 5.1, Recommendation II.4).

Understanding Financial Markets

While being an important channel for financing of the real economy, securities markets have also been a key channel for risk transmission, particularly due to the increasing reliance on market-based financing. Therefore, there is consensus on

accounts. The annual reports are coordinated collections of data aggregated to a global level to allow for the analysis of global trends and risks in the shadow banking system. For the first time, the 2015 report introduced a new measure of shadow banking based on the economic functions of non-bank financial entities focusing only on those nonbank fi-nancial institutions that are involved in significant maturity and liquidity transformation or leverage, and are part of a credit-intermediation chain. This allows policymakers to better focus on the potential risks shadow banking entities may pose. Based on this measure, the global assets of finan-cial entities classified as shadow banking reached $45 trillion at the end of 2016.

Regarding oversight and regulation of the shadow bank-ing system, among the topics covered is that of risks in the securities lending and repurchase markets. The crisis pointed out that short-term deposit-like funding of nonbank entities can easily lead to “runs” in the market if confidence is lost. The use of these collateralized funding (secured financing) techniques can exacerbate such runs and boost leverage, es-pecially when asset prices are buoyant and margins and hair-cuts on secured financing are low. Therefore, the FSB initiated work to collect and aggregate data on securities fi-nancing markets that is now incorporated in DGI-2 (Appen-dix 5.1, Recommendation II.5). Preparation for official data collection and aggregation started in 2018.

Monitoring the Global Systemically Important Financial Institutions

Due to the significance of global systemically important fi-nancial institutions (G-SIFIs) in spreading shocks across borders, and the potential effects of their failure for the global financial system, several measures were taken to im-prove the resilience of these institutions to limit the moral-hazard effects. Among these measures were the identification of G-SIBs in 2011 by the Basel Committee on Banking Su-pervision and the introduction of additional loss-absorbency measures for such institutions. Having better data on the bilateral linkages of these institutions as well as their expo-sures to and funding dependencies on national financial sys-tems was seen as an important prerequisite to understanding the risks associated with these institutions. To this end, the work to construct a data template for G-SIFIs as recom-mended by the DGI focused initially on G-SIBs.

The end product of this exercise, which was led by the FSB, in close consultation with the IMF, is a set of unique data templates bringing together consistent, granular infor-mation on G-SIBs that is useful for both microprudential and macroprudential analysis. Collection of data started with information on G-SIBs’ bilateral linkages as well as some aggregate information based on the institution-level data underlying the consolidated IBS and will continue with the collection of information on G-SIBs’ exposures to and lending from key economies with the granularity of a combination of sector, instrument, currency, and maturity.

12 The sharing of reports based on G-SIBs’ data with international finan-cial institutions (BIS, FSB, and IMF) under strict confidentiality condi-tions has been established.



objectives of this reform include reporting of OTC derivative contracts to trade repositories and trading of all standardized contracts on exchanges or electronic trading platforms, where appropriate, with clearing through central counterparties. Non-centrally-cleared contracts are subject to higher capital requirements. This regulatory initiative to clear OTC deriva-tives through central clearing allows for more standardiza-tion of reporting and aggregation both for regulatory and financial-data purposes. In turn, these developments are also increasing interest in the quality of reporting as well as the consistency among the already existing data collections.

How Does the DGI Address the Surveillance Agenda?

As noted elsewhere in this chapter, in the wake of the 2014 TSR, the IMF Managing Director published an Action Plan for Strengthening Surveillance. Among the actions to be taken was that “The Fund will revive and adapt the balance sheet approach to facilitate a more in-depth analysis of the impact of shocks and their transmission across sectors.” This responded to a call from outside experts David Li and Paul Tucker in their external study for the 2014 TSR on risks and spillovers (see IMF 2014b).

Sectoral Analysis

Even though the 2007–08 crisis emerged in the financial sector, given its intermediary role, the problems in the finan-cial sector also affected other economic sectors. Therefore, analysis of balance sheet exposures is essential, given the in-creasingly interconnected global economy. As it is pointed out in the IMF TSR 2014b, the use of balance sheets to iden-tify sources of vulnerability and the transmission of shocks could have helped detect risks associated with European banks’ reliance on US wholesale funding to finance struc-tured products.

In June 2015, the IMF set out the way forward in a paper for the IMF Executive Board on Balance Sheet Analysis in Surveillance (IMF 2015a). Sectoral accounts and balance sheet data are essential, including from-whom to-whom data, in providing the context for an assessment of the links between the real economy and financial sectors. The sectoral balance sheets of the SNA are seen as the overarching frame-work for balance sheet analysis as the IMF Executive Board paper makes clear. Further, the paper sets out a data frame-work for such analysis (IMF 2015a, 23).

Putting the sectoral balance sheets of the SNA in a policy context, the IMF has developed a balance sheet approach, which compiles all of the main balance sheets in an economy using aggregate data by sector. The balance sheet approach is based on the same conceptual principles as the sectoral accounts, providing information on a from-whom-to-whom basis with an additional focus on vulnerabilities arising from maturity and currency mismatches as well as the capital structure of economic sectors. While currently not that

the importance of better information on these markets in order to understand the diversification of funding sources and the exposures of both issuers and creditors, including the nonfinancial sector. Long important in advanced econo-mies, there is evidence of growing security issuance in emerging market economies as the composition of corporate debt has been shifting away from loans and toward bonds (IMF 2015c). At the same time, it is estimated that over the past decade, domestic debt securities markets in emerging market economies have increased from around one third of emerging market economies’ GDP to about one half (Hat-tori and Takáts 2015).

The DGI has addressed this growing policy interest in securities markets by providing conceptual advice through the publication of a Handbook on Securities Statistics (see BIS, ECB, and IMF 2015a) prepared jointly by the BIS, the Euro-pean Central Bank, and the IMF. The DGI has also ad-dressed this interest by fostering improvements in securities statistics through encouraging G20 economies to report to the BIS database on securities statistics.

Since 2009, the number of economies that report regular and consistent securities statistics to the BIS has increased significantly. Moreover, even though the levels of sophistica-tion of national statistical frameworks are diverse among G20 economies, these countries increasingly recognize the importance of having granular information on these markets, and hence are considering building security-by-security data-bases.13 Data on securities issuance (and holdings) are also an input into national accounts, balance of payments, and gov-ernment finance statistics. The initial focus of the DGI-2 is to improve data on issuance of debt securities with key informa-tion on the markets, sectors, currency, maturity, and interest rate. Consistent information on holdings of debt securities and from-whom-to-whom data is considered a longer-term objective (Appendix 5.1, Recommendation II.7).

The need to bring light to the opaqueness of the OTC derivatives markets is also a focus of the DGI. In DGI-1 CDS data were expanded both in detail and country cover-age, and regular reporting of the expanded datasets was im-plemented. All economies with significant CDS markets report CDS data to the BIS survey, including more detail on the type and geography of counterparties as well as the un-derlying instrument (BIS 2017).

In DGI-2, there is recognition of the need to improve data on OTC derivative markets more broadly (Appendix 5.1, Recommendation II.6). In September 2009, G20 leaders agreed to a comprehensive reform agenda to improve trans-parency in OTC derivatives markets, mitigate systemic risk, and protect against market abuse. They asked the FSB and its relevant members to assess its implementation regularly. The

13 The BIS database on international debt securities, which has been devel-oped based on granular information, allowing for the parallel identifi-cation of the residency and nationality of debt securities issuers, is an example of a security-by-security database constructed at the interna-tional level.



Further, the IMF Executive Board began addressing the problem of government finance statistics in 2010, and reaf-firmed this in 2013 by requiring the inclusion in staff reports of key elements of the GFSM presentation (see IMF 2013b). The paper also confirmed the intention to establish a Gov-ernment Finance Statistics Advisory Committee to support the implementation of GFSM and advise on emerging fiscal data issues. The advisory committee met for the first time in March 2015 and agreed to support development of the GFSM 2014 with a number of practical recommendations (see IMF 2015e).

Despite these developments at the international level, progress has lagged behind other recommendations. This is due to factors such as the lack of coverage of state and local governments, the fact that Government Finance Statistics in many countries are not institutionally well established, and, in some instances, the reluctance of authorities to use statis-tical techniques to fill the data gaps.14 To this end, the DGI-2 remains aimed at addressing the gaps in government finance statistics (Appendix 5.1, Recommendation II.15).

Within the same context, information on the debt levels of the public sector, particularly general government, is cru-cial to assess the fiscal soundness of government. Under the DGI, the World Bank, the OECD, and the IMF launched a quarterly public sector debt statistics database in 2010 to promote standardized reporting by countries. However, the scope of sector and instrument coverage can differ signifi-cantly across countries. This is highly relevant because of the close analytical and policy interest in measures such as gross debt to GDP. If a country “only” covers debt securities and loans for the budgetary central government, comparing these data with a country that covers all debt liabilities in-cluding accounts payable and pension obligations for the general government will clearly not be comparing like with like (see Dippelsman, Dziobek, and Gutiérrez Mangas 2012). As a consequence, supported by DGI-2, the World Bank’s public debt database is moving to a presentation of instrument and sectoral coverage on a matrix basis—pre-senting varying levels of sector and instrument coverage from the narrowest to the broadest (IMF 2015f) (Appendix 5.1, Recommendation II.16).15

Understanding Cross-Border Financial Interconnections

The crisis emphasized the fact that it is not possible to isolate the problems in a single financial system, as shocks propa-gate rapidly across the financial systems. Indeed, since 2010 the IMF has been identifying jurisdictions with systemically

many economies compile from-whom-to-whom balance sheet data, balance sheet approach data can be compiled from the IMF’s Standardized Report Forms, IIP, and gov-ernment balance sheet data—a more limited set of data than needed to compile the sectoral accounts.

The DGI-2 recommendations address key data gaps that act as a constraint on a full-fledged balance sheet analysis. The DGI recommends addressing such gaps through im-proving G20 economies’ dissemination of sectoral accounts and balance sheets, building on the 2008 System of National Accounts, including for the nonfinancial corporate and household sectors (Annex 5.1, Recommendation II.8). Given the multifaceted character of the datasets, implementation of this recommendation is challenging, and progress has been slow. However, all G20 economies agree on the impor-tance of having such information and have plans in place to make it happen.

In a world of capital flow liberalization and fewer credit con-straints, widening distributions of income, consumption, sav-ing, and wealth can lead to potential financial vulnerabilities even if the aggregate data look reassuring. Indeed, the impor-tance of good distributional data for households has become increasingly apparent over recent years as policy interest in in-equality has increased in both advanced and developing econo-mies in recent decades (see IMF’s Work on Income Inequality at http://www.imf.org/external/np/fad/inequality/index.htm). DGI-2 focuses on the compilation of distributional information (such as information by income quintiles) to complement ag-gregate figures, consistent with national accounts (Appendix 5.1, Recommendation II.9). To this end, the Organisation for Economic Co-operation and Development (OECD) has car-ried out conceptual work on distributional information focus-ing on (1) linking national accounts with distributional information (microdata and macrodata), summarized in two OECD publications (OECD 2013a; OECD 2013b), and (2) the provision of an improved conceptual alignment of income, consumption, and wealth in microsurveys, including a further enhancement of wealth definitions.

Analysis of Fiscal Condition

Significant data gaps also exist in the area of government fi-nance statistics. The support provided by many national au-thorities to the financial sector following the global financial crisis, along with the onset of recession and fiscal stimulus programs to support demand, led to increases in fiscal defi-cits and government debt. However, consistent and compa-rable fiscal data across the G20 economies was lacking, hampering cross-country analysis. Further, monitoring the trends in the fiscal positions of governments was often limited by a lack of frequent and timely harmonized data, including a lack of accrual-based data.

At the international level, there has been significant prog-ress in support of the compilation of government finance statistics. Conceptual work has included publication of the Government Finance Statistics Manual 2014 (GFSM 2014).

14 In addition, government finance statistics are not always consistent with the relevant data in the national accounts despite the harmonization of international standards across statistical domains.

15 The BIS has published a dataset on credit to the general government sector for 26 advanced and 14 emerging market economies (Dembier-mont and others 2015).


http://www.imf.org/external/np/fad/inequality/index.htm


try and sector of banks’ counterparties, in particular non-bank financial institutions. The enhancements to the IBS will provide users with a more comprehensive picture of the size and scope of internationally active banks’ activities. Hence, it will enable a better analysis of the sources and uses of funds and of the importance of international business for banks of different nationalities (Avdjiev, McGuire, and Wooldridge 2015).

The CPIS started to be collected on a semiannual fre-quency from June 2013 with a dissemination lag of less than nine months (see IMF 2011). In 2018, all G20 economies reported CPIS data to the IMF on a semiannual basis. Fur-ther, there is growing interest in understanding the sector allocation of holders and issuers, in order to match the sec-toral analysis in the domestic accounts. At the end of 2017, 16 G20 economies reported the sector of holder data, while among the enhancements made in the DGI were new tables for collecting information on the sector of the issuer of secu-rities, also crossed with sector of the holder of securities. DGI-2 maintains the focus on improving CPIS reporting of G20 economies.

The IMF’s CDIS complements the CPIS and IBS for an analysis of cross-border interconnectedness, as it provides in-formation on direct investment positions broken down by net equity and net debt. The CDIS was brought under the DGI umbrella in its second phase with an aim toward improving the quality of G20 economies direct investment position statistics, both inward and outward.

Foreign currency risk is an important element of an anal-ysis of cross-border interconnections. To this end, particular attention was given, including by the G20 FMCBG (see BIS, FSB, and IMF 2015b) to the improvement of foreign cur-rency exposure information given the potential spillover ef-fects of wealth transfers triggered by sharp movements in exchange rates. Within this context, the IMF focused on improving the compilation of foreign currency exposures data across its statistical domains, particularly through the IIP. The BIS contributes to the analysis of foreign currency exposures through its international debt securities, and its enhanced IBS, which provides the basis for deriving a more detailed picture of internationally active banks’ balance sheets and thus measuring potential currency mismatches more accurately.

The global financial crisis also revealed the need to under-stand better the cross-border foreign currency exposures of nonfinancial corporations. The FSB Committee on the Global Financial System16 identified the gaps in information on foreign currency exposures of corporations in a joint workshop in 2014, reporting the outcomes to the G20 FM-CBG. In addition, the FSB conducted in 2014 a peer review of the trade repository reporting of OTC derivatives, covering all types of OTC derivatives, including foreign currency

important financial sectors based on a set of relevant and transparent criteria, including size and interconnectedness. Within this identification framework, cross-border intercon-nectedness is considered an important complementary mea-sure to the size of the economy: it captures the systemic risk that can arise through direct and indirect interlinkages among financial sectors in the global financial system (that is, the risk that failure or malfunction of a national financial system may have severe repercussions on other countries or on overall systemic stability) (IMF 2010).

The 2014 TSR summed up the issue succinctly in its Ex-ecutive Summary: “Risks and spillovers remain first-order issues for the world economy and should be central to Fund surveillance. Recent reforms have made surveillance more risk-based, helping to better capture global interconnections. Experience so far also points to the need to build a deeper understanding of how risks map across countries, and how spillovers can quickly spread across sectors to expose domes-tic vulnerabilities” (IMF 2014d).

Four existing datasets that include key information on cross-country financial linkages are the IIP, BIS IBS, IMF CPIS, and IMF Coordinated Direct Investment Survey (CDIS). Together, these datasets provide a comprehensive picture of cross-border financial interconnections. This pic-ture is especially relevant for policymakers, as financial con-nections strengthen across borders and domestic conditions are affected by financial developments in other economies to which they are closely linked financially. DGI-2 focuses on improving the availability and cross-country comparability of these datasets (Appendix 5.1, Recommendations II.10, 11, 12, and 13).

The well-known IIP is a key data source to understanding the linkages between the domestic economy and the rest of the world by providing information on both external assets and liabilities of the economy with a detailed instrument breakdown. However, the crisis revealed the need for cur-rency and more detailed sector breakdowns, particularly for the other financial corporations sector. Consequently, as part of the DGI, the IIP was enhanced to support these pol-icy needs. Significant progress has also been made in ensur-ing regular reporting of IIPs along with an increase in frequency of reporting from annual to quarterly. By the end of 2017 all G20 economies reported quarterly IIP data.

The IBS have been a key source of data for many decades, providing information on aggregate assets and liabilities of internationally active banking systems on a quarterly fre-quency. The CPIS data, while on an annual frequency, pro-vided significant insights into portfolio investment assets. That said, both datasets had limitations in terms of country coverage and granularity. CPIS also needed to be improved in terms of frequency and timeliness. To this end, the DGI supported the enhancements in these datasets.

The IBS was enhanced through (1) expanded coverage of banks’ balance sheets to also include domestic positions in complement to cross-border activities, and (2) improved granularity of data by collecting more information on coun-

16 The Committee on Global Financial System is the BIS committee of central banks overseeing the collection of the IBS statistics.



ers, given the link with household consumption and the need to monitor asset prices in an environment of accom-modative monetary policies. Consequently, most of the G20 economies have been increasing their efforts to develop good statistics on real estate prices. Residential real estate price is one of the FSIs that is prescribed for adherents to the SDDS Plus. Good real estate price indices can also support the mea-surement of nonfinancial assets in the sectoral accounts.

Commercial property price indices are at a less-developed stage, both conceptually and in terms of availability of data. However, there is a financial stability interest in the dissemi-nation of this data for monitoring asset bubbles. This is be-cause commercial property is used for banks’ collateralized lending; and commercial property price indices data are im-portant for the valuation of securitized assets. Therefore, work needs to be done to enhance the methodological guid-ance that is being developed and to encourage the provision of available data to the BIS for public dissemination.

DGI-2 Recommendations II.17 and 18 address the devel-opment of property place indices.

The Need to Better Communicate Statistics

The DGI facilitated a new tier of the IMF’s data standards (SDDS Plus) mainly intended for economies with systemi-cally important financial systems to guide IMF member countries on the provision of economic and financial data to the public in support of domestic and international financial stability. Economies adhering to the SDDS Plus are expected to disseminate data in nine categories covering four macro-economic sectors—the real sector, the fiscal sector, the financial sector, and the external sector—all of which were largely drawn from the DGI-1 recommendations. Given the common areas of focus, adhering to the SDDS Plus and im-plementation of DGI recommendations would contribute to each other. When this book went to press, 18 countries were in adherence with the SDDS Plus.

The IMF also continues to consider the needs of emerg-ing markets and low-income countries. The SDDS aims to enhance the availability of timely and comprehensive statis-tics and thereby contribute to the pursuit of sound macro-economic policies and the improved functioning of financial markets. In May 2015, the GDDS was enhanced (e-GDDS) to assist countries with relatively less-developed statistical capacity. The emphasis on data dissemination in the e-GDDS will support transparency, encourage statistical de-velopment, and help create strong synergies between data dissemination and surveillance. IMF will continue working with member economies, including through capacity devel-opment activities, to ensure the availability and dissemina-tion of information.

There was also a need to improve the communication of official statistics, as in some instances users were not fully aware of the available data series to address critical policy issues. As part of the DGI, the Principal Global Indicators (PGI) website (http://www.principalglobalindicators.org),

derivatives and other instruments that create foreign cur-rency exposures (FSB 2015a). This work was also reported to the G20. All in all, the DGI-2 supports this work on foreign currency exposures through more explicit incorporation of data on foreign currency exposures in the recommendations (that is, IIP, cross-border exposures of nonbank corporations, and securities statistics).

The increase in foreign exchange derivative exposures of nonfinancial corporations through off-shore entities has been an area of concern, particularly for emerging market economies, as authorities were unaware of the transactions recorded outside their jurisdictions. The DGI framework is contributing to shedding light on this broader area of cross-border exposures and intragroup funding by nonfinancial corporations through their off-shore subsidiaries (Appendix 5.1, Recommendation II.14). Conceptual guidance was pro-vided to clarify nationality, group, and consolidation con-cepts in DGI-1 (IAG 2015), and going forward, BIS and IMF will continue to improve information on nonfinancial corporations’ cross-border exposures, mainly drawing on their existing data collections. The OECD will also contrib-ute through the development of a framework that links its multinational enterprises data with its foreign direct investment data.

Monitoring the Property Markets

Residential and commercial property price indices are im-portant for the detection and monitoring of asset price bub-bles, the compilation of estimates of household and corporate wealth and capital formation, and the assessment of the broader financial stability implications. The relevance of property prices was stressed at the Second IMF Statistical Forum (see Silver 2014), while work at the BIS has high-lighted the importance of asset price developments— especially property prices—in driving the so-called financial cycle (see Borio and Drehmann 2009; and Dembiermont 2015).

However, the availability and international comparability of this data was limited before the global financial crisis. As part of the DGI-1, conceptual guidance was provided through the publication of Handbook on Residential Property Price Indices,17 and in 2010 the BIS started to disseminate real estate price statistics on its website. Currently, most G20 countries provide data even though the data provided are often at a development stage and more work is needed to ensure consistency and international comparability.

Over recent years, the importance of good real estate price statistics has become increasingly clear to policymak-

17 See Handbook on Residential Property Prices Indices (European Union and others 2013), which was jointly supported by Eurostat and the In-ternational Labour Organization, IMF, OECD, UN, and World Bank.


http://www.principalglobalindicators.org


legal frameworks to support microdata access while preserv-ing confidentiality.

Broader Implications of the DGI

Even though the focus of the DGI has been the G20 econo-mies, it involves a wider range of economies as it builds on widely accepted international statistical frameworks. In particular, the methodological guidance provided as a result of the DGI is for all IMF member economies. Moreover, most recommendations of the DGI build on various statisti-cal initiatives that involve a larger group of economies. Furthermore, as these non-G20-member economies see the merit in this initiative for their own policy work, even without any higher level policy push such as the G-20 FM-CBG, they are working toward implementation of the DGI recommendations.

Evidently, improved quality of information worldwide is es-sential to ensure a complete assessment of global macro-finan-cial linkages. With this in mind, a reference note (IMF 2015b) to the IMF paper Balance Sheet Analysis in Fund Surveillance (IMF 2015a), provided a full listing of available balance-sheet-related macro datasets, including their relevance for surveil-lance, and a summary of data availability for each IMF member. Many of the datasets referenced were those covered by DGI-2. Further, the IMF staff provides increasing support to member countries for the compilation of these datasets through technical assistance and training (see IMF 2016).

DGI and Stress Testing

As explained in IMF 2012, “Stress testing is a technique that measures the vulnerability of a portfolio, an institution, or an entire financial system under different hypothetical events or scenarios. It is a quantitative “what if” exercise, estimating what would happen to capital, profits, cash flows, etc. of individual financial firms or the system as a whole if certain risks were to materialize” (IMF 2012). By increasing the availability, consistency, and comparability of data rele-vant for financial stability analysis, the DGI supports the use of stress tests to assess a broad range of vulnerabilities (for example, credit risk, market risk, and foreign exchange risk), with increased reliability and more flexibility in the poten-tial range of scenarios tested.

More specifically, the work under the DGI to increase the granularity of data will improve the sensitivity of the stress testing exercises and allow for assessments of vulnerabilities of specific components of the financial system (for example, specific sectors and instruments). Further, improved avail-ability, quality, and consistency of institutions’ own data sets (for example, data on global systemically important banks) could be considered a key input to bottom-up stress tests.

Starting with the financial sector, stress testing exercises using systematically collected and disseminated FSI data can flag issues for follow-up, not only on individual institutions (microprudential) but also on aggregate financial systems

hosted by the IMF, was launched in 2009 as a joint undertaking of the IAG with the aim of facilitating the monitoring of economic and financial developments. The PGI website includes data for the G20 economies and non-G20 members that have systemically important financial sectors and are subject to a five-year mandatory FSAP.18 The PGI website was significantly enhanced as part of the DGI in terms of coverage and timeliness. It currently offers access to an online database with user-selected longer runs of his-torical data presented in comparable units of measure (growth rates, index numbers, and percent of GDP). Work to strengthen the PGI further will continue in DGI-2 (Appendix 5.1, Recommendation II.19).

As new risks emerge and relationships between institu-tions, sectors, and countries get more complex, the granular-ity of data needed to assess those risks become relevant. This brings along the need for compilers of data to collect infor-mation at the micro level to help meet user demands. DGI recommendations (that is, G-SIBs data, IBS, CPIS) support the need for more granular data.19 On the other hand, the increasing granularity of data also raises challenges for sharing such information either within economies, across borders, and with international agencies, due to confidenti-ality concerns. This potentially limits the broader benefits of new or existing data collections including some under the DGI. While this is not an easy problem to tackle, DGI-2 focuses on the issue of confidentiality by encouraging the G20 economies to increase the sharing and accessibility of granular data, if needed, by revisiting existing confidential-ity constraints (Appendix 5.1, Recommendation II.20). Ad-dressing the confidentiality constraints and how they can be overcome as part of the DGI-2 would be a positive step forward.

It is also worth noting that the private sector is working toward improving bank disclosures. To this end, the En-hanced Disclosure Task Force, a private sector group of fi-nancial institutions established by the FSB, released a report in 2012 that included seven fundamental principles for enhancing the risk disclosures of banks.20

The need for granular and real-time data to understand in-terconnectedness and spillovers across countries and institu-tions was also emphasized at the Third IMF Statistical Forum held in Frankfurt in November 2015 (http://www.imf.org /external/np/seminars/eng/2015/statsforum/). The participants urged statistical agencies and policymakers to establish new

18 Currently, 29 IMF member countries are subject to mandatory finan-cial stability assessments, while for the rest of the membership the “FSAP” is voluntary.

19 Micro, granular data sources can also enhance the accuracy and level of details of “traditional” macro statistics. See Tissot 2015.

20 The report focused on disclosures in risk governance and risk manage-ment, capital adequacy and liquidity, funding, market risk, credit risk and other risks. See FSB 2012.





might lose value during an external shock, and how the loss in value is funded. Indeed, the increased focus on foreign currency data in the DGI-2 further supports such analysis through stress testing

Finally, the extent of the trade and financial links across borders turned out to be deeper and more firmly established than most were aware of during the global financial crisis, including through cross-border bond holdings and banking links and through trade supply lines.21 Various datasets in-cluded in the DGI-2 support the stress testing of cross- border financial links, including the IBS, CPIS, and CDIS data.

Besides the few examples provided previously, implemen-tation of the DGI-2, in general, would contribute to the stress testing exercises by improving their sensitivity, increas-ing their scope, and allowing for more flexibility in the po-tential range of scenarios tested.

3. CONCLUSIONThe DGI could be considered an overarching initiative cov-ering a wide range of statistical frameworks that are inter-linked in support of the common goal of understanding financial markets and instruments and shedding light on interconnectedness.

Making available a comprehensive set of information, as intended by the DGI, in a standardized, frequent, and timely manner is not an easy task, especially while also ensuring that the available data are reliable, of high quality, and properly reflect the changing economic circumstances. This cannot happen overnight, but over time and with a global effort, it is possible. It also requires high-level support, including re-sources to be secured. Since the global financial crisis, signifi-cant efforts have been made by all relevant parties to ensure that the policymakers have access to DGI-related informa-tion as a key component of their toolbox. To be able to reap the benefits of the investments made in the DGI, it is impor-tant to maintain the pace of work and continue coordination among all players in the global economy.

(macroprudential). For instance, the nonperforming-loans-to-total-loans ratio, one of the core FSIs, is a widely used indicator for the assessment of potential shocks on credit quality.

Sharp changes in property prices are among the key risk factors used in macroeconomic stress scenarios. The increas-ing availability of comprehensive data on property prices, along with other data being enhanced through the DGI, in-cluding sectoral accounts, would therefore allow for a broader range of stress testing exercises to assess the impact of real estate price changes on the both financial sector and the economy more broadly.

For a more aggregate economy-wide perspective, the avail-ability of sectoral accounts data allows for the construction of many indicators of vulnerability including household debt to income, and the relative shift in activity of financial institu-tions, such as from banks to nonbank financials, while also providing a tool for analyzing the link between the real and fi-nancial economies. Analysis through stress tests of the poten-tial impact of various scenarios, including regulatory reform on financial sector activity, and on the economy more broadly, can be enhanced by a comprehensive set of sectoral accounts.

The often close ties between the government and the financial sector can potentially lead to a negative sovereign-bank feedback loop: financial sector problems can lead to bailouts by government, and the financial sector can have large exposures to governments (such as through security holdings, often encouraged by regulatory policy) that are facing fiscal stresses. The more detailed and comprehensive data for government accounts and securities holdings being promoted under DGI-2 would support stress testing scenar-ios to monitor such potential negative feedbacks.

With regard to the external sector, the channel through which exposures to the rest of the world will primarily affect the domestic economy is the IIP, be it through transactions; current or financial account; valuation; changes in market prices and exchange rates; or other flows, such as debt write-offs. IIP data—with its integrated system of stocks and flows—can be utilized along with sectoral accounts for stress tests to analyze which sectors might gain and which sectors

21 The IMF Direction of Trade data can be used to analyze cross-border trade linkages.


Appendix 5.1.DGI-2 Recommendations

Recommendation II.1: Mandate of the Data Gaps Initiative (DGI)

The G20 economies to regularly compile comparable and high-quality economic and financial statistics in accordance with interna-tional standards and disseminate such statistics in a timely manner. The Inter-Agency Group on Economic and Financial Statistics (IAG) to coordinate and monitor the implementation of the DGI recommendations, and promote the Principal Global Indicators (PGI) website as a global reference database. Staff of the Financial Stability Board (FSB) and IMF to provide annual updates on progress to G20 finance ministers and central bank governors.

MONITORING RISKS IN THE FINANCIAL SECTORRecommendation II.2: Financial Soundness Indicators

The G20 economies to report the seven financial soundness indicators (FSIs) expected from Special Data Dissemination Standard Plus adherent economies, on a quarterly frequency. G20 economies are encouraged to report the core and expanded lists of FSIs, with a particular focus on other (nonbank) financial corporations. The IMF to coordinate the work and monitor progress.

Recommendation II.3: Concentration and Distribution Measures

The IMF to investigate the possibility of regular collection of concentration and distribution measures for FSIs. G20 economies to sup-port the work of the IMF.

Recommendation II.4: Data for Global Systemically Important Financial Institutions

The G20 economies to support the International Data Hub at the Bank for International Settlements (BIS) to ensure the regular col-lection and appropriate sharing of data about global systemically important banks. In addition, the FSB, in close consultation with the IMF and relevant supervisory bodies, to investigate the possibility of a common data template for systemically important nonbank fi-nancial institutions starting with insurance companies. This work will take due account of the confidentiality and legal issues.

Recommendation II.5: Shadow Banking

The G20 economies to enhance data collection on the shadow banking system by contributing to the FSB monitoring process, including through the provision of sectoral accounts data. FSB to work on further improvements of the conceptual framework and developing standards and processes for collecting and aggregating consistent data at the global level.

Recommendation II.6: Derivatives

BIS to review the derivatives data collected for the International Banking Statistics (IBS) and the semiannual over-the-counter de-rivatives statistics survey, and the FSB to develop a mechanism to aggregate and share at global level over-the-counter derivatives data from trade repositories. The G20 economies to support this work as appropriate.

Recommendation II.7: Securities Statistics

G20 economies to provide on a quarterly frequency debt securities issuance data to the BIS consistent with the Handbook on Security Statistics starting with sector, currency, type of interest rate, original maturity and, if feasible, market of issuance. Reporting of hold-ings of debt securities and the sectoral from-whom-to-whom data prescribed for Special Data Dissemination Standard Plus adherent economies would be a longer-term objective. BIS, with the assistance of the Working Group on Securities Databases, to monitor regu-lar collection and consistency of debt securities data.



VULNERABILITIES, INTERCONNECTIONS, AND SPILLOVERSRecommendation II.8: Sectoral Accounts

The G20 economies to compile and disseminate, on a quarterly and annual frequency, sectoral accounts flows and balance sheet data, based on the internationally agreed template, including data for the other (nonbank) financial corporations sector, and develop from-whom-to-whom matrices for both transactions and stocks to support balance sheet analysis. The IAG, in collaboration with the Inter-secretariat Working Group on National Accounts, to encourage and monitor the progress by G20 economies.

Recommendation II.9: Household Distributional Information

The IAG, in close collaboration with the G20 economies, to encourage the production and dissemination of distributional information on income, consumption, saving, and wealth, for the household sector. The Organisation for Economic Co-operation and Development (OECD) to coordinate the work in close cooperation with Eurostat and European Central Bank.

Recommendation II.10: International Investment Position (IIP)

The G20 economies to provide quarterly IIP data to the IMF, consistent with the Balance of Payments and International Invest-ment Position Manual, sixth edition, and including the enhancements, such as the currency composition and separate identification of other (nonbank) financial corporations, introduced in that manual. IMF to monitor reporting and the consistency of IIP data, and consider separate identification of nonfinancial corporations, in collaboration with IMF Committee on Balance of Payments Statistics.

Recommendation II.11: International Banking Statistics (IBS)

G20 economies to provide enhanced BIS international banking statistics. BIS to work with all reporting countries to close gaps in the reporting of IBS, to review options for improving the consistency between the consolidated IBS and supervisory data, and to support efforts to make data more widely available.

Recommendation II.12: Coordinated Portfolio Investment Survey

G20 economies to provide, on a semiannual frequency, data for the IMF Coordinated Portfolio Investment Survey, including the sec-tor of holder table and, preferably, also the sector of nonresident issuer table. IMF to monitor the regular reporting and consistency of data, to continue to improve the coverage of significant financial centers, and to investigate the possibility of quarterly reporting.

Recommendation II.13: Coordinated Direct Investment Survey (CDIS)

G20 economies to participate in and improve their reporting of the IMF Coordinated Direct Investment Survey, both inward and outward direct investment. IMF to monitor the progress.

Recommendation II.14: Cross-Border Exposures of Nonbank Corporations

The IAG to improve the consistency and dissemination of data on nonbank corporations’ cross-border exposures, including those through foreign affiliates and intragroup funding, to better analyze the risks and vulnerabilities arising from such exposures including foreign currency mismatches. The work will draw on existing data collections by the BIS and the IMF, and on the development of the OECD framework for foreign direct investment. The G20 economies to support the work of the IAG.

Recommendation II.15: Government Finance Statistics

The G20 economies to disseminate quarterly general government data consistent with the Government Finance Statistics Manual 2014 (GFSM 2014). Adoption of accrual accounting by the G20 economies is encouraged. The IMF to monitor the regular reporting and dissemination of timely, comparable, and high-quality government finance data.



Recommendation II.16: Public Sector Debt Statistics

The G20 economies to provide comprehensive general government debt data with broad instrument coverage to the World Bank/IMF/OECD Public Sector Debt Database. The World Bank to coordinate the work.

Recommendation II.17: Residential Property Prices

The G20 economies to publish residential property price indices consistent with the Handbook on Residential Property Price Indices and supply these data to the relevant international organizations, including the BIS, Eurostat, and OECD. The IAG in collaboration with the Inter-Secretariat Working Group on Price Statistics to work on a set of common headline residential property price indices; encouraging the production of long time series; developing a list of other housing-related indicators; and disseminating the headline residential property price data via the PGI website.

Recommendation II.18: Commercial Property Prices

The IAG in collaboration with the Inter-Secretariat Working Group on Price Statistics to enhance the methodological guidance on the compilation of Commercial Property Price Indices and encourage dissemination of data on commercial property prices via the BIS website.

COMMUNICATION OF OFFICIAL STATISTICSRecommendation II.19: International Data Cooperation and Communication

The IAG to foster improved international data cooperation among international organizations and support timely standardized transmis-sion of data through internationally agreed formats (for example, Statistical Data and Metadata eXchange) to reduce the burden on reporting economies, and promote outreach to users. The IAG to continue to work with G20 economies to present timely, consistent na-tional data on the PGI website and on the websites of participating international organizations.

Recommendation II.20: Promotion of Data Sharing

The IAG and G20 economies to promote and encourage the exchange of data and metadata among and within G20 economies, and with international agencies, to improve the quality (for example, consistency) of data, and availability for policy use. The G20 econo-mies are also encouraged to increase the sharing and accessibility of granular data, if needed, by revisiting existing confidentiality constraints.



Financial Stability Board (FSB). 2012. EDTF Principles and Rec-ommendations for Enhancing the Risk Disclosures of Banks. Report of the Enhanced Disclosure Task Force. Basel: Finan-cial Stability Board. http://www.fsb.org/2012/10/r_121029/.

———. 2013. Strengthening Oversight and Regulation of the Shadow Banking: An Overview of Policy Recommendations. Progress Re-port to G20 Ministers and Governors. Basel: Financial Stabil-ity Board. http://www.fsb.org/2013/08/an-overview-of-policy -recommendations-for-shadow-banking/.

———. 2014. FSB Data Gaps Initiative—A Common Data Tem-plate for Global Systemically Important Banks. Basel: Financial Stability Board. http://www.fsb.org/2014/05/r_140506/.

———. 2015a. Corporate Funding Structures and Incentives. Basel: Financial Stability Board. http://www.fsb.org/2015/09/corporate -funding-structures-and-incentives/.

———. 2015b. Global Shadow Banking Monitoring Report. Basel: Financial Stability Board. http://www.fsb.org/2015/11/global -shadow-banking-monitoring-report-2015/.

———. 2015c. Transforming Shadow Banking into Resilient Market-Based Finance. Basel: Financial Stability Board. http://www .fsb.org/2015/11/transforming-shadow-banking-into-resilient -market-based-finance-an-overview-of-progress/.

——— and International Monetary Fund (IMF). 2009. The Fi-nancial Crisis and Information Gaps. Report to the G20 Finance Ministers and Central Bank Governors. Basel: Financial Stabil-ity Board, and Washington, DC: International Monetary Fund. http://www.fsb.org/2009/10/r_091029/.

———. 2015. Sixth Progress Report on the Implementation of the G20 Data Gaps Initiative. Basel: Financial Stability Board, and Washington, DC: International Monetary Fund.

Group of Twenty (G20). 2008. “Declaration of the Summit on Financial Markets and the World Economy.” Washington, DC, November 15. https://georgewbush-whitehouse.archives.gov /news/releases/2008/11/20081115-1.html.

Hattori Masazumi, and Takáts Előd. 2015. “The Role of Debt Se-curities Markets.” BIS Paper 83, Bank for International Settle-ments, Basel, Switzerland.

Heath, Robert. 2013. “Why Are the G20 Data Gaps Initiative and the SDDS Plus Relevant for Financial Stability Analysis?” IMF Working Paper 13/6, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2016 /12/31/Why-are-the-G-20-Data-Gaps-Initiative-and-the -SDDS-Plus-Relevant-for-Financial-Stability-40227.

———. 2015. “What Has Capital Flow Liberalization Meant for Economic and Financial Statistics?” IMF Working Paper 15/88, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 /What-has-Capital-Liberalization-Meant-for-Economic-and -Financial-Statistics-42842.

———, and Evrim Bese Goksu. 2016. “G20 Data Gaps Initiative II: Meeting the Policy Challenge.” IMF Working Paper 16/43, Inter-national Monetary Fund, Washington, DC. https://www.imf.org /en/Publications/WP/Issues/2016/12/31/G-20-Data-Gaps -Initiative-II-Meeting-the-Policy-Challenge-43760.

Inter-Agency Group on Economic and Financial Statistics (IAG). 2015. “Consolidation and Corporate Groups: An Overview of Methodological and Practical Issues.” IAG Reference Document, Bank for International Settlements, Basel, Switzerland.

International Association of Insurance Supervisors (IAIS). 2011. “In-surance and Financial Stability.” International Association of In-surance Supervisors, Basel, Switzerland. https://www.iaisweb.org /page/supervisory-material/other-supervisory-papers-and-reports.

REFERENCESAvdjiev, Stephen, Patrick McGuire, and Philip Wooldridge. “En-

hanced Data to Analyse International Banking.” 2015. BIS Quarterly Review (September). https://www.bis.org/publ/qtrpdf /r_qt1509f.htm.

Borio, Claudio, and Mathias Drehmann. 2009. “Assessing the Risk of Banking Crises—Revisited.” BIS Quarterly Review (March). https://www.bis.org/publ/qtrpdf/r_qt0903e.htm.

Bank for International Settlements (BIS). 2017. Semiannual OTC Derivatives Statistics. https://www.bis.org/statistics/derstats .htm?m=6%7C32%7C71.

———, European Central Bank (ECB), and International Mone-tary Fund (IMF). 2015a. Handbook on Securities Statistics. Washington, DC: International Monetary Fund. https://www .bis.org/publ/othp23.htm.

———, Financial Stability Board (FSB), and International Mon-etary Fund (IMF). 2015b. Work on Foreign Currency Exposures. Report to G20 Economies. Basel: Bank for International Set-tlements and Financial Stability Board, and Washington, DC: International Monetary Fund.

Crowley, Joseph, Plapa Koukpamou, Elena Loukoianova, and An-dre Mialou. 2016. “Pilot Project on Concentration and Distri-bution Measures for a Selected Set of Financial Soundness Indicators.” IMF Working Paper 16/26, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/Pilot-Project-on-Concentration -and-Distribution-Measures-for-a-Selected-Set-of-Financial -43706.

Dembiermont, Christian. 2015. “Measuring Property Prices: The BIS Contribution.” Paper presented at the 2015 ISI World Statistics Congress, Rio de Janeiro, Brazil. https://www.bis.org/ifc/events/ifc _isi_2015.htm.

———, Michela Scatigna, Robert Szemere, and Bruno Tissot. 2015. “A New Database on General Government Debt.” BIS Quarterly Review (September). https://www.bis.org/publ/qtrpdf /r_qt1509g.htm.

Dippelsman, Robert, Claudia Dziobek, and Carlos A. Gutiérrez Mangas. 2012. “What Lies Beneath: The Statistical Definition of Public Sector Debt.” IMF Staff Discussion Note 12/09, International Monetary Fund, Washington, DC. https://www .imf.org/external/pubs/cat/longres.aspx?sk=26101.

Errico, Luca, Artak Harutyunyan, Elena Loukoianova, Richard Wal-ton, Yevgeniya Korniyenko, Goran Amidžić, Hanan AbuShanab, and Hyun Song Shin. 2014. “Mapping the Shadow Banking Sys-tem Through a Global Flow of Funds Analysis.” IMF Working Paper 14/10, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 /Mapping-the-Shadow-Banking-System-Through-a-Global -Flow-of-Funds-Analysis-41273.

European Union, International Labour Organization, International Monetary Fund, Organisation for Economic Co-operation and Development, United Nations, Economic Commission for Europe, and World Bank. 2013. “Handbook on Residential Property Prices Indices.” Eurostat Methodologies and Working Papers, European Union, Luxembourg. https://ec.europa.eu /eurostat/web/products-manuals-and-guidelines/-/KS-RA -12-022.

Fender, Ingo, and Patrick McGuire. 2010. “Bank Structure, Funding Risk and the Transmission of Shocks across Coun-tries: Concepts and Measurement.” BIS Quarterly Review (September). http://www.bis.org/publ/qtrpdf/r_qt1009h.htm.


http://www.fsb.org/2012/10/r_121029/.

http://www.fsb.org/2013/08/an-overview-of-policy-recommendations-for-shadow-banking/.

http://www.fsb.org/2013/08/an-overview-of-policy-recommendations-for-shadow-banking/.

http://www.fsb.org/2014/05/r_140506/.

http://www.fsb.org/2015/09/corporate-funding-structures-and-incentives/.

http://www.fsb.org/2015/09/corporate-funding-structures-and-incentives/.

http://www.fsb.org/2015/11/global-shadow-banking-monitoring-report-2015/.

http://www.fsb.org/2015/11/global-shadow-banking-monitoring-report-2015/.

http://www.fsb.org/2015/11/transforming-shadow-banking-into-resilient-market-based-finance-an-overview-of-progress/



http://www.fsb.org/2009/10/r_091029/.

https://georgewbush-whitehouse.archives.gov/news/releases/2008/11/20081115-1.html

https://georgewbush-whitehouse.archives.gov/news/releases/2008/11/20081115-1.html

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Why-are-the-G-20-Data-Gaps-Initiative-and-the-SDDS-Plus-Relevant-for-Financial-Stability-40227.



https://www.imf.org/en/Publications/WP/Issues/2016/12/31/What-has-Capital-Liberalization-Meant-for-Economic-and-Financial-Statistics-42842.



https://www.imf.org/en/Publications/WP/Issues/2016/12/31/G-20-Data-Gaps-Initiative-II-Meeting-the-Policy-Challenge-43760



https://www.iaisweb.org/page/supervisory-material/other-supervisory-papers-and-reports

https://www.iaisweb.org/page/supervisory-material/other-supervisory-papers-and-reports

https://www.bis.org/publ/qtrpdf/r_qt1509f.htm

https://www.bis.org/publ/qtrpdf/r_qt1509f.htm


https://www.bis.org/statistics/derstats.htm?m=6%7C32%7C71.

https://www.bis.org/statistics/derstats.htm?m=6%7C32%7C71.

https://www.bis.org/publ/othp23.htm.

https://www.bis.org/publ/othp23.htm.

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Pilot-Project-on-Concentration-and-Distribution-Measures-for-a-Selected-Set-of-Financial-43706




https://www.bis.org/ifc/events/ifc_isi_2015.htm.,

https://www.bis.org/ifc/events/ifc_isi_2015.htm.,

https://www.bis.org/publ/qtrpdf/r_qt1509g.htm

https://www.bis.org/publ/qtrpdf/r_qt1509g.htm

https://www.imf.org/external/pubs/cat/longres.aspx?sk=26101

https://www.imf.org/external/pubs/cat/longres.aspx?sk=26101

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Mapping-the-Shadow-Banking-System-Through-a-Global-Flow-of-Funds-Analysis-41273



https://ec.europa.eu/eurostat/web/products-manuals-and-guidelines/-/KS-RA-12-022



http://www.bis.org/publ/qtrpdf/r_qt1009h.htm


———. 2015b. “Balance Sheet Analysis in Fund Surveillance— Reference Note.” IMF Policy Paper, Washington, DC. https://www .imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Balance -Sheet-Analysis-in-Fund-Surveillance-Reference-Note-PP4968.

———. 2015c. Global Financial Stability Report: Vulnerabilities, Legacies, and Policy Challenges, Chapter 3. Washington, DC, Oc-tober. https://www.imf.org/en/Publications/GFSR/Issues/2016 /12/31/Vulnerabilities-Legacies-and-Policy-Challenges.

———. 2015d. Proceedings of the Third IMF Statistical Forum: Official Statistics to Support Evidence-Based Policymaking, Frankfurt am Main, Germany, November. http://www.imf .org/external/np/seminars/eng/2015/statsforum/.

———. 2015e. Proceedings of the Meeting of the IMF Govern-ment Finance Statistics Advisory Committee, Washington, DC, March. http://www.imf.org/external/pubs/ft/gfs/gfsac /meetings/2015.

———. 2015f. “Progress with Globally Comparable Public Sector Debt Statistics.” Paper presented at the IMF Government Fi-nance Statistics Advisory Committee, Washington, DC, March. https://www.imf.org/external/pubs/ft/gfs/gfsac/meetings/2015/.

———. 2015g. “Rethinking Macro Policy III: Progress or Confu-sion?” IMF Seminar Series, Washington, DC, April. http://www.imf.org/external/np/seminars/eng/2015/macro3/.

———. 2016. Statistics Department at a Glance. Washington, DC: International Monetary Fund.

Navajas, Matias Costa, and Aaron Thegeya. 2013. “Financial Soundness Indicators and Banking Crises.” IMF Working Pa-per 13/263, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 /Financial-Soundness-Indicators-and-Banking-Crises-41168.

Organisation for Economic Co-operation and Development (OECD). 2013a. OECD Framework for Statistics on the Distribution of House-hold Income, Consumption and Wealth. Paris: OECD Publishing. http://www.oecd.org/statistics/framework-for-statistics-on-the -distribution-of-household-income-consumption-and-wealth -9789264194830-en.htm.

———. 2013b. OECD Guidelines for Micro Statistics on Household Wealth. Paris: OECD Publishing. http://www.oecd.org/statistics /guidelines-for-micro-statistics-on-household-wealth.htm.

Silver, Mick. 2014. “Real Estate Prices: Availability, Importance, and New Developments.” Paper presented at the Second IMF Statistical Forum: Statistics for Policymaking, Washington, DC, November. https://www.imf.org/external/np/seminars/eng /2014/statsforum/.

Tissot, Bruno. 2015. “Closing Information Gaps at the Global Level: What Micro Data Can Bring.” Irving Fisher Committee Workshop: Combining Micro and Macro Statistical Data for Financial Stability Analysis: Experiences, Opportunities and Challenges. Bank for International Settlements, Basel, Decem-ber 14–15. https://www.bis.org/ifc/publ/ifcb41.htm.

United Nations. 1953. “A System of National Accounts and Sup-porting Tables.” Studies in Methods. No. 2. https://unstats .un.org/unsd/nationalaccount/hsna.asp.

———. 2013. “Global Systemically Important Insurers: Initial Assessment Methodology.” International Association of Insurance Supervisors, Basel, Switzerland. https://www.iaisweb.org/page /supervisory-material/financial-stability-and-macroprudential-policy -and-surveillance//f ile/34256/final-g-siis-policy-measures -18-july-2013.

International Monetary Fund (IMF). 1948. Balance of Payments Manual. Washington, DC: International Monetary Fund.

———. 2010. “IMF Expanding Surveillance to Require Manda-tory Financial Stability Assessments of Countries with Systemi-cally Important Financial Sectors.” IMF Press Release No. 10/357. https://www.imf.org/external/np/sec/pr/2010/pr10357 .htm.

———. 2011. “Enhancements to the Coordinated Portfolio Invest-ment Survey.” Paper presented at the 24th Meeting of the IMF Committee on Balance of Payments Statistics, Moscow, Russia. https://www.imf.org/external/pubs/ft/bop/2011/24.htm.

———. 2012. “Macro-Financial Stress Testing—Principles and Practices.” IMF Policy Paper, Washington, DC. https://www .imf.org/en/Publications/Policy-Papers/Issues/2016/12/31 /Macrof inancial-Stress-Testing-Principles-and-Practices -PP4702.

———. 2013a. Modifications to the Current List of Financial Soundness Indicators. Washington, DC: International Mone-tary Fund. https://www.imf.org/en/Publications/Policy-Papers /Issues/2016/12/31/Modif ications-to-the-Current-List-of -Financial-Soundness-Indicators-PP4832.

———. 2013b. Review of the Implementation of Government Fi-nance Statistics to Strengthen Fiscal Analysis. Washington, DC: International Monetary Fund. https://www.imf.org/external /pubs/ft/gfs/manual/comp.htm.

———. 2014a. Proceedings of the Second IMF Statistical Forum: Statistics for Policymaking, Washington, DC, November. http://www.imf.org/external/np/seminars/eng/2014/statsforum/.

———. 2014b. “Triennial Surveillance Review—External Study, Risks and Spillovers.” IMF Policy Paper, Washington, DC. https://www.imf.org/en/Publications/Policy-Papers/Issues /2016/12/31/2014-Triennial-Surveillance-Review-External -Study-Risks-and-Spillovers-PP4902.

———. 2014c. “Triennial Surveillance Review—Managing Direc-tor’s Action Plan for Strengthening Surveillance.” IMF Policy Paper, Washington, DC. https://www.imf.org/en/Publications/Po l i c y - Pa p e r s / I s s u e s /2 016 /12 /31/2 014 -Tr i e n n i a l -Surveillance-Review-Managing-Directors-Action-Plan-for -Strengthening-PP4924.

———. 2014d. “Triennial Surveillance Review—Overview Paper.” IMP Policy Paper, Washington, DC. https://www.imf.org/en /Publ icat ions/Pol ic y-Papers/ I s sues/2016/12/31/2014 -Triennial-Surveillance-Review-Overview-Paper-PP4897.

———. 2015a. “Balance Sheet Analysis in Fund Surveillance.” IMF Policy Paper, Washington, DC. https://www.imf.org/en /Publications/Policy-Papers/Issues/2016/12/31/Balance-Sheet -Analysis-in-Fund-Surveillance-PP4963.


https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Balance-Sheet-Analysis-in-Fund-Surveillance-Reference-Note-PP4968.



https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Vulnerabilities-Legacies-and-Policy-Challenges.

https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Vulnerabilities-Legacies-and-Policy-Challenges.

http://www.imf.org/external/np/seminars/eng/2015/statsforum/.


http://www.imf.org/external/pubs/ft/gfs/gfsac/meetings/2015.

http://www.imf.org/external/pubs/ft/gfs/gfsac/meetings/2015.

https://www.imf.org/external/pubs/ft/gfs/gfsac/meetings/2015/.

http://www.imf.org/external/np/seminars/eng/2015/macro3/.

http://www.imf.org/external/np/seminars/eng/2015/macro3/.

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Financial-Soundness-Indicators-and-Banking-Crises-41168

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Financial-Soundness-Indicators-and-Banking-Crises-41168

http://www.oecd.org/statistics/framework-for-statistics-on-the-distribution-of-household-income-consumption-and-wealth-9789264194830-en.htm.



http://www.oecd.org/statistics/guidelines-for-micro-statistics-on-household-wealth.htm

http://www.oecd.org/statistics/guidelines-for-micro-statistics-on-household-wealth.htm

https://www.imf.org/external/np/seminars/eng/2014/statsforum/

https://www.imf.org/external/np/seminars/eng/2014/statsforum/

https://www.bis.org/ifc/publ/ifcb41.htm

https://unstats.un.org/unsd/nationalaccount/hsna.asp.

https://unstats.un.org/unsd/nationalaccount/hsna.asp.

https://www.iaisweb.org/page/supervisory-material/financial-stability-and-macroprudential-policy-and-surveillance//file/34256/final-g-siis-policy-measures-18-july-2013




https://www.imf.org/external/np/sec/pr/2010/pr10357.htm.

https://www.imf.org/external/np/sec/pr/2010/pr10357.htm.

https://www.imf.org/external/pubs/ft/bop/2011/24.htm.





https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Modifications-to-the-Current-List-of-Financial-Soundness-Indicators-PP4832.



https://www.imf.org/external/pubs/ft/gfs/manual/comp.htm.

https://www.imf.org/external/pubs/ft/gfs/manual/comp.htm.



https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/2014-Triennial-Surveillance-Review-External-Study-Risks-and-Spillovers-PP4902.



https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/2014-Triennial-Surveillance-Review-Managing-Directors-Action-Plan-for-Strengthening-PP4924.




https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/2014-Triennial-Surveillance-Review-Overview-Paper-PP4897.



https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Balance-Sheet-Analysis-in-Fund-Surveillance-PP4963



CHAPTER 6

Ring-Fencing and Consolidated Banks’ Stress Tests

EUGENIO CERUTTI • CHRISTIAN SCHMIEDER

The recent crisis has spurred the use of bank stress tests as a crisis management and early warning tool. However, a weakness is that current stress tests are based on consolidated balance sheets, and thus omit potential risks embedded in banking groups’ geographical structures by assum-

ing that capital and liquidity are available wherever they are needed within the group. This chapter presents a framework to integrate ring-fencing and regulatory differences (for example, minimum capital requirements) into cross-border bank stress tests. Case studies show how some forms of ring-fencing—that is, home or host regulators limiting flows of capital and income within a group—could significantly increase banks’ capital needs.

This chapter was previously published as “Ring Fencing and Consolidated Banks’ Stress Tests” in the Journal of Financial Stability (Cerutti and Schmieder 2014). The authors would like to thank Stijn Claessens, Laura Kodres, Liliana Schumacher, Jorge A. Chan-Lau, two anonymous reviewers, and participants of IMF Research Department seminars and a Monetary and Capital Markets Department Senior Staff Meeting for their helpful comments and suggestions during the different stages of this project. The views expressed in this chapter are those of the authors and should not be attributed to the IMF or the Bank for International Settlements, their Boards, or their management.1 See Foglia 2009; Ong and Čihák 2010; and Borio, Drehmann, and Tsatsaronis 2014 for recent discussions on stress test limitations, and Cerutti, Claes-

sens, and McGuire 2014 for the data challenges in the context of systemic risk analysis for global banking.2 Note that this geographical perspective of ring-fencing is different from the activity restrictions embedded in the Volcker Rule (section 619 of the Dodd

Frank Act), which restricts deposit-taking banks from engaging in certain types of activities (for example, proprietary trading).

provides evidence, given existing data limitations, that a stress test approach using both consolidated and unconsoli-dated balance sheet data is necessary and relevant due to the potential implications of “ring-fencing,” defined as partially or fully limiting the ability of cross-border banking groups to reallocate funds from subsidiaries with excess capital and/or liquidity to those in need of capital and/or liquidity.2

In this context, a straightforward conceptual approach for how unconsolidated and consolidated balance sheet data can be combined is developed in order to take into account the risks potentially embedded in banking groups’ geo-graphical structures. In addition, this chapter presents evi-dence on the cost of (partial) full ring-fencing for the largest European banks. The ring-fencing approach builds on Ce-rutti and others 2010, which was the first paper measuring the potential important impact of different degrees of ring-fencing through simulations on banks’ subsidiaries in Emerging Europe. Although ring-fencing is currently exten-sively discussed in the policy arena, very little empirical work has been done in this area. Two related exceptions are

1. INTRODUCTIONThe global financial crisis and subsequent events revealed weaknesses in the stress testing exercises (and other types of risk, stability, and early warning exercises) carried out by the public and private sectors. Subsequently, the toolbox has been bolstered in recent years, in ways that have addressed a number of methodological issues (that is, the sophistication of stress tests in technical terms, inclusion of liquidity and contagion, and so on) and scenario-related considerations (that is, the severity and scope of shocks). Nevertheless, sev-eral weaknesses and challenges remain, many of them re-lated to the lack of adequate data, especially from a cross-border context.1

This chapter focuses on the limitations of carrying out stress tests using consolidated banking groups’ balance sheets and income statements, especially when stress testing international banking groups. Stress tests run at the group level using “only” consolidated data do not take into account the possibility that home or host regulators might limit or even fully block flows within banking groups. This chapter


Ring-Fencing and Consolidated Banks’ Stress Tests102

Cerutti and others 2010 documented anecdotal evidence that bank regulators in Croatia, Poland, and Turkey limited the distribution of net income by subsidiaries of foreign banks despite relatively strong bank fundamentals in 2009. Anecdotal evidence also points to potential episodes in core European Union countries. For example, German supervi-sors tried to clamp down Unicredit’s increased borrowing from its German subsidiary in 2012, and British regulators barred the UK arm of Banco Santander from transferring funds to its Spanish parent.6

This anecdotal evidence is likely underestimating the prevalence of ring-fencing restrictions since banks reported that supervisors also made use of informal moral suasion or unpublished Basel II/III prudential requirements (EBRD 2013). It becomes even more difficult to assess the level of ring-fencing, since the degree to which host country authori-ties ring-fence their foreign affiliates as well as the capacity of foreign banking groups for working around host country restrictions are a function of the severity of the crisis. A for-eign bank could work around host country supervisors’ re-strictions by selling their subsidiaries (or part of them) to domestic banks or other foreign banks, provided that host supervisors agree to the specific deal.7

Ring-fencing considerations are mainly relevant for the large cross-border banks around the globe, such as the large EU banking groups with significant diversified geographical structure in terms of assets, liabilities, and profits. This fea-ture and the fact that the June 2011 EBA stress test released detailed consolidated data motivate this chapter’s focus on large European banks. Our estimates, using projections based on banks’ 2010 data from the EBA exercise, indicate up to 3 percentage points of additional Core Tier 1 capital needs for banks under very strict forms of ring-fencing from all supervisors of their outside EU subsidiaries, and up to 2.4 percent if simulations are circumscribed to the countries for which anecdotal ring-fencing evidence has been docu-mented. It should be noted, however, that our numerical results do not necessarily reflect current circumstances—running stress tests to test banks’ current resilience was not the purpose of this chapter. Many banks have since raised

Schoenmaker and Siegmann (2014) and van Lelyveld and Spaltro (2011), who estimated the cost associated with fore-cast-based burden sharing agreements, but not the impact of ring-fencing on banks’ capital buffers. From a theoretical angle, the literature (see Freixas 2003; and Dell’Ariccia and Marquez 2006) has also highlighted how the lack of fore-cast-based coordination across national regulators can lead to an inefficient outcome due to the underprovision of finan-cial stability at a global level (that is, like in a standard public good problem).3

The European stress tests run by the European Banking Authority (EBA) in 2010–11 as well as the stress tests run by US authorities in 2009 and 2012 are good examples of the mentioned progress in, and limitations of, recent stress tests.4

Both exercises came up with a series of conceptual improve-ments, which increased stress test coverage and sophistica-tion. However, these stress tests were conducted using consolidated banks’ balance sheets, thus assuming that re-sources available at one location within a group could im-mediately be used in another location. This assumption is in line with the literature on multinational banks’ internal capital markets. For example, De Haas and van Lelyveld (2010) and Cetorelli and Goldberg (2012a, 2012b) have highlighted that global banks have (to some extent) been able to reallocate funds across locations in response to their relative needs.

While this assumption is likely to be always valid within countries and likely to be valid more generally in closely in-tegrated jurisdictions with similar rules (for example, the European Union), evidence from crisis periods has shown that this is also frequently not the case. This is because (at least) some degree of ring-fencing by host supervisors is likely at such times. While foreign subsidiaries do not neces-sarily need to suffer explicit discrimination during periods of stress, regulators have imposed additional unilateral restric-tions covering all (or many) banks in their jurisdictions in order to safeguard national financial stability, effectively limiting international banking groups’ ability to reallocate resources within the group.5 Moreover, in some cases, host supervisors seem to have explicitly targeted foreign banks. The European Bank for Reconstruction and Development’s compilation of unilateral financial sector measures during the crisis mentions that supervisory measures in Albania, Poland, and the Czech Republic directly covered parent banks’ operations (for example, restricting transactions be-tween liquid foreign subsidiaries in the Czech Republic and their foreign parent banks) (EBRD 2013). In addition,

3 See Hardy and Nieto 2011, for example, who analyze the beneficial ways to come up with a joint design of prudential supervision and de-posit guarantee regulations.

4 See EBA 2011, Federal Reserve 2012, and Appendix 6.1 for more details.

5 For example, regulators in Albania, Bulgaria, the Czech Republic, Hungary, Poland, Romania, Serbia, and the Slovak Republic issued unilateral financial sector measures to safeguard national financial sta-bility (EBRD 2013).

6 Ring-fencing restrictions can also originate from home-country super-visors. For example, Austria’s regulators—worried about increasing im-pairments in Eastern Europe—pushed Austrian banks to reduce lending in that region, eliciting protests from countries such as Hun-gary and Romania. Austria later partly rescinded the rules (for more details see “Turmoil Frays Ties Across Continent,” Wall Street Journal, May 31, 2012).

7 Nevertheless, this is sometimes not easy during crises. For example, de-spite the fact that Dexia, National Bank of Greece and Banco Comer-cial Portugues announced in late 2010–11 their initial intentions to sell their respective Denizbank, Finansbank, and Millennium profitable subsidiaries in Turkey and Poland, only Dexia was able to do so in June 2012 since the global financial crisis forced potential buyers to focus on increasing their capital ratios rather than expanding. There can also be an adverse market reaction to potential fire sales of affiliates during cri-ses, due to the classical lemon problem linked to valuation uncertainty of assets.


Eugenio Cerutti and Christian Schmieder 103

capital and some of them have changed their geographical structure (for example, by selling subsidiaries).

The magnitude of the estimated adjustments is compara-ble to Basel III’s proposed (up to 2½ percent) Core Tier 1 capital adequacy capital surcharge to be applied on global systemically important banks (G-SIBs).8 The G-SIB sur-charge would broadly cover the estimated additional capital needed for the most affected banks under the full ring-fenc-ing scenario from non-EU countries. However, the Basel III calibration of the level of additional G-SIB capital sur-charges is not meant to be a buffer against ring-fencing, yet, in conceptual terms, the classification of G-SIBs into differ-ent risk buckets is based, among the five factors, on cross-jurisdictional activity and bank size. These two factors, especially the cross-border activity, are also important in ex-plaining the results of our analysis. The establishment of a potential (regulatory) capital buffer that would explicitly ac-count for ring-fencing would require further simulations based on different ring-fencing assumptions (for example, different severity levels of stress, and the potential impact of ring-fencing behavior in different parts of the world), and have to include all major cross-border banks—US banks, Swiss banks, and so on.

The more comprehensive stress testing approach for banking groups proposed herein (compared with existing stress tests) helps to gain a better understanding of the po-tential vulnerabilities of a banking group, both as a whole and relative to its parts (for example, a specific subsidiary or region). A more detailed view of banking group vulnerabili-ties would be informative to both banks and supervisors in assessing how a potential financial shock in one country or region could affect specific subsidiaries and the group as a whole. In this context, the proposed approach and illustrative results also highlight the need for more international coop-eration. The establishment of a credible framework for the resolution of cross-border banking groups would help to avoid unilateral and likely more costly solutions (IMF 2010).9 Without such a credible international cooperation (which in-cludes burden-sharing arrangements), larger capital buffers for international banks that take into account the potential impact of ring-fencing under certain plausible scenarios could be a second-best option, but establishing a level playing field will be a great challenge.

This chapter is organized as follows: Section 2 presents a concept on how stress testing of banking groups can be un-dertaken using both unconsolidated and consolidated bal-ance sheet data. Accordingly, Section 3 provides case studies of the resulting impact. Section 4 concludes and discusses policy implications.

2. A CONCEPT TO COMBINE UNCONSOLIDATED AND CONSOLIDATED STRESS TESTSDespite various improvements in stress-testing techniques in the aftermath of the global financial crisis and the ongoing sovereign debt crisis, important challenges remain to be ad-dressed. The current implicit assumption of full mobility of capital and liquidity within banking groups is far from triv-ial. This section presents a simple conceptual framework on how to take into account banking groups’ geographical di-versification, how to overcome current data gaps, and how to reflect different degrees of ring-fencing in the scenarios.

Conceptual Framework for a Combined Unconsolidated/Consolidated Approach

As depicted in Figure 6.1, most of the current stress tests fol-low a “traditional” consolidated top-down approach using consolidated bank balance sheet data. In the “best” case, this consolidated top-down approach partially takes into ac-count the geographical structure of banking groups’ business through granular macroeconomic scenarios, differentiating between specific different countries and regions when pro-jecting income, impairments, and other relevant solvency parameters under stress (for example, total assets, risk-weighted assets, dividend payouts, and so on). This was the case of the 2011 EBA stress tests that came up with a global macroeconomic scenario, which was then broken down into country-specific trajectories of GDP growth, inflation, and unemployment rates for each of the 27 EU member coun-tries, and scenarios for the major non-EU countries. See Ap-pendix 6.1 for more details.

Nevertheless, this is not enough to fully take into account the geographical structure of banking groups, which could include several independent foreign-incorporated bank sub-sidiaries.10 Such geographical characteristics can be incorpo-rated into the analysis using a bottom-up approach based on both unconsolidated and consolidated data, which we also refer to as a group-structure approach.

This approach explicitly treats foreign subsidiaries and the parent bank (at unconsolidated terms) as single entities,

8 Basel III uses an indicator-based approach to group G-SIBs into four categories of systemic importance. The selected indicators reflect the size of banks, their interconnectedness, the lack of readily available sub-stitutes or financial institution infrastructure for the services they pro-vide, their global (cross-jurisdictional) activity, and their complexity. For more details see Basel Committee on Banking Supervision 2011 and Financial Stability Board 2012.

9 See Eisenbeis and Kaufman 2008 and Krimminger 2008 for further discussions on the challenges in achieving international coordination and effective implementation of cross-border resolution. Avgouleas, Goodhart, and Schoenmaker (2013) highlight the opportunity to use bank resolution plans and living wills to introduce a consistent legal regime for the resolution of systemically important institutions.

10 In some countries, especially in Latin America, the distinction between branches and subsidiaries is blurring. Foreign branches have been re-quired to adhere to the domestic regulation (for example, move capital to the country). See Cerutti, Dell’Arriccia, and Martinez Peria 2007 for more details.



stress tests given that banks have access to granular data of their own businesses, while the global macroeconomic sce-nario (and with country-specific impact) could be established by the authorities.

Best-practice stress tests (using granular scenarios for dif-ferent geographical regions and business lines), such as the 2011 EBA stress test, will, in principle, similarly allow for the establishment of a detailed view of the risk profile of banks (as the scenarios are broken down to country and exposure data, respectively).11 However, taking a group consolidated perspec-tive does not allow for a simulation of the impact of ring- fencing, which can be a key shortcoming for some banks.

Data Issues

A key question is the availability of data to run a meaningful group-structure stress test.

The best solution is to get actual bank data (for parent banks and individual subsidiaries). This option is only avail-able to banks themselves (for example, for their own internal stress tests and as part of bottom-up stress tests) and to su-pervisors (provided that the reported supervisory data is suf-ficiently granular).12

A second-best solution is to put together publicly avail-able data on bank subsidiaries, a burdensome job, but one that substantially facilitates the understanding of banks’

and then aggregates the outcome of stress tests for single en-tities based on a set of assumptions regarding potential de-grees of ring-fencing (by defining how banks’ incomes, excess capital, and/or liquidity can be transferred within the banking group). Take the example on the right-hand side of Figure 6.1 of a parent bank with three subsidiaries, which shows the detailed view provided by the group-structure ap-proach compared with the consolidated approach (shown on the left-hand side). Two advantages of the group-structure approach emerge:

1. Setup: The group-structure approach allows for a more granular view of stability, but is more complex and burdensome.

2. Scenario design: The group-structure approach ex-plicitly forces one to come up with a consistent global macroeconomic scenario that is broken down into scenarios for the subsidiaries. The scenarios could also include one or more scenarios for the de-gree of ring-fencing (see green-colored box in Figure 6.1). Together, this determines how banks’ business structures affect their solvency and/or liquidity posi-tions under specific circumstances.

The key benefits of using the group-structure approach are that it forces stress testers to go through a number of thought processes, namely (1) the design of consistent global scenarios, broken down to different entities (often into sce-narios for specific countries); (2) the development of an un-derstanding of the banks’ business models and how they can be affected by different shocks and policy actions, respec-tively; and (3) the uncovering of potential policy challenges that could emerge under certain circumstances, which could contribute to contingency planning.

In addition, unconsolidated group-wide tests can also re-veal the relative soundness of specific entities within a group (for example, a subsidiary), which can inform policymakers (see IMF 2012a, 2012b, for more details). Bottom-up stress tests run by banks are a natural candidate for group- structure

Macroeconomic Scenario Macroeconomic Scenario

Outcome for the Group

OutcomeParent Bank

OutcomeSub 1

OutcomeSub 2

OutcomeSub 3

Outcome for the Group

Alternatives in terms of ring-fencing(subject to bank-specific min ratios)

Traditional ConsolidatedStress Test

Unconsolidated Stress Test Taking into Account Group Structure

Source: Authors.Note: min = minimum; sub = subsidiary.

Figure 6.1 Conceptual Difference between “Traditional” Stress Tests and Stress Tests Taking into Account Group Structures

11 The European stress tests (EBA 2011) simulated country-specific stress levels, for example, and applied them to the exposures banks have in certain countries. It did not consider the risks of ring-fencing.

12 The availability of supervisory data would allow for a fully fledged im-plementation of the bottom-up unconsolidated approach. Subsidiaries’ capital and liquidity calculations could account for potential differences (for example, risk weights) between the group level and its subsidiaries. For instance, even though host country sovereign debt could have a very low risk/high liquidity weight for a subsidiary due to local regulations, the group-level risk weight of the same claim on the local government could have a much higher risk weight following an international credit agency assessment.



case is especially relevant when assuming full ring-fencing. In such circumstances, the macroeconomic environment that is being modeled would be important, with significant repercussions on the asset value of banks.

3. QUANTIFYING THE POTENTIAL BIAS OF NOT USING A COMBINED APPROACHUsing figures from the June 2011 EBA stress test, this sec-tion illustrates, through simulations, the potential implica-tions of explicitly considering ring-fencing in stress tests. As noted above, the examples are meant for illustrative purposes and do not necessarily reflect the current situation of banks, many of which have raised capital and changed their legal and/or geographical structures since then. Nevertheless, the magnitude of the impact can provide benchmarks of what cross-border banks might encounter under highly adverse conditions.

The baseline, which implies no ring-fencing, is given by EBA figures. In each of the case studies, the impact of two adjustments—partial and full ring-fencing—is calculated to account for the possibility that supervisors ring-fence foreign deposit-taking affiliates (that is, bank subsidiaries) in their jurisdictions.14 Facing potential spillovers from an external shock (for example, an EU shock), a host country supervisor could opt to ring-fence its foreign bank affiliates by impos-ing restrictions on intragroup cross-border transfers. The case studies cover 51 banking groups included in the EBA June 2011 stress test (see more details in Appendix 6.2).

The first case study simulates the impact of all non-EU supervisors ring-fencing their foreign affiliates of EU banks, which reflects the broadest possible ring-fencing coverage to be triggered by all regulators outside EU banking common market rules. However, anecdotal evidence documented that some form of ring-fencing was also exercised by specific countries within the EU. Thus, the second set of simulations focuses on the impact of ring-fencing by specific supervisors, with a view to analyzing different degrees of ring-fencing by: (1) supervisors for whom there is anecdotal evidence of ring-fencing; and (2) the ones who host important subsidiaries (in terms of the host country’s total bank assets).

Ring-Fencing of EU Banks by Non-EU Supervisors

For the first case study, the chapter puts together a picture of the EU banking groups’ operations outside the EU by com-paring publicly available balance sheet data of deposit-taking affiliates with the consolidated statements at the group level.

business. For example, proxy data could be the residual of group data (that is, the implied parent) and data for major subsidiaries.

Ultimately, the burden of putting together data and es-tablishing a meaningful stress test has to be compared with the potential gain in insights one would expect. For a bank-ing system that is mostly domestically owned and where banks do not have noteworthy foreign subsidiaries, the value added would be rather limited. This would be different in systems that host international banks.

Introducing Ring-Fencing

After computing the impact of the macro-financial stress test assumptions for each subsidiary and the parent—in terms of solvency and/or liquidity, and taking into account the different national regulations (for example, minimum regulatory capital ratios)—the next step is to define a set of assumptions (that is, scenarios) as to how banks’ incomes, excess capital and/or liquidity can be transferred within the banking group under stress. We propose the following three alternatives to simulate potential ring-fencing, but the final choice of the scenario should be tailored to the objectives of the stress test at hand, and (broadly) reflect the severity of the scenario:13

• No ring-fencing assumes that the parent bank’s in-come, as well as subsidiaries’ excess liquidity and ex-cess capital buffers can be used to cover capital shortfall in any of the subsidiaries. With the excep-tion that the group-structure approach would have to take into account different country regulations (for example, minimum regulatory capital ratios), this case is basically covered by the implicit assump-tions in the current consolidated approaches.

• Partial ring-fencing assumes that the parent bank’s in-come and/or only subsidiaries’ income and/or excess liquidity, but not excess capital, can be re-allocated within a group.

• Full ring-fencing assumes that no transfers between any of the group’s affiliates (including from the par-ent bank to subsidiaries) can take place. This case of full ring-fencing would correspond with a strict stand-alone subsidization approach—where a cross-border banking group is set up as a network of fully self-sufficient national subsidiaries.

Besides ring-fencing in a narrower sense, the scenarios could also consider limits on any income, capital, or liquid-ity generated by sales of assets. Parent banks could try to overcome some form of ring-fencing by local authorities through selling part or all of the subsidiaries’ assets. This

14 Due to data limitations, the cross-country example presented here is not a full-fledged unconsolidated type of stress test exercise as proposed in the previous section. For example, EBA projections of capitalization of net income are only available at the banking group aggregate, without including the breakdowns among the different parts of the group.

13 Cerutti and others (2010) included a fourth alternative: near-complete ring-fencing, which assumed that only transfers from the parent to any of the subsidiaries are allowed as an intermediate alternative between partial and full ring-fencing. Since we are not interested in the results of the stress tests for a particular subsidiary, we did not include it.



There is also no uniform pattern across banking groups if we consider size (total assets). The expected positive correla-tion between size of banks and the share of both net income and capital outside the EU is present, but not necessarily as strong as expected (0.42 and 0.39, respectively). Moreover, two clear business models are highlighted by the data (Figure 6.2). Traditional, retailed-oriented banks, such as Santander, Banco Bilbao Vizcaya Argentaria, S.A. (BBVA), HSBC, Unicredit, Raiffeisen, and the National Bank of Greece (NBG) have a more substantial presence outside the EU (that is, about 20 percent or more of net income) than the remaining banking groups have, independent of their size. And the remaining banks, which include some with a sizable investment business such as Deutsche Bank, display an over-all lower share of deposit-taking affiliates’ assets, net income, and capital outside the EU.16

The EBA tests provided two projections, a baseline sce-nario (based on macroeconomic baseline projections by the EU Commission at that time) and an adverse double-dip scenario, both for a two-year horizon (2011–12) based on end-of-2010 balance sheets. Based on the concept outlined in Section 2, the chapter estimates the impact of both partial ring-fencing and full ring-fencing for both EBA scenarios, assuming that all non-EU supervisors ring-fence their spe-cific affiliates. As the impact of ring-fencing is compared with the outcome based on the consolidated approach (that is, actual EBA results), and does not focus on the resulting capital levels, the choice of the EBA stress levels is less rele-vant. In the case studies presented in this section, we gener-ally refer to the impact under EBA baseline conditions.

As shown in Appendix 6.2, large European banking groups have up to 25 percent of their assets, 43 percent of capital, and 27 percent of net income outside the EU.15 However, there is substantial heterogeneity in terms of the geographi-cal structure of large EU banks. About two-thirds of the banking groups operate almost exclusively within the EU and the portion of their income (average of 2006–10), capi-tal, and assets outside the EU below 5 percent. On the other hand, about 10 banking groups have a large presence outside the EU, many of them having more than 20 percent of their net income and capital located in bank-deposit affiliates out-side the EU. In addition, the shares of capital and income outside the EU tend to be generally larger than the shares of assets, indicating that operations outside the EU require more capital than within the EU—which could be driven by higher capital requirements as well as by higher risks—and are also more profitable.

The share of non-EU activities through deposit-taking af-filiates does not seem to be closely related to the home coun-try of the parent bank. Although, for example, the large Spanish banks included in the sample have a large share of their activities outside the EU, the rest of the Spanish bank-ing system does not have a large international presence. Aus-trian, Greek, Italian, and UK banks also display a large heterogeneity in this respect.

0

30

Total Banking Group Assets (USD billion)

Net I

ncom

e Ou

tsid

e Eu

rope

0

50

Total Banking Group Assets (USD billion)

Capi

tal O

utsi

de E

urop

e

Net Income Outside EU(Percent)

Capital Outside EU(Percent)

5

10

15

20

25

0

500

1000

1500

2000

2500

3000

3500

10

20

30

40

0

500

1000

1500

2000

2500

3000

3500

Santander HSBC

NBG

Raiffeisen

BBVAUnicredit

Barclays

DeutscheBNP Paribas

ING Bank

Credit Agricole

Santander

HSBC

RBS

BBVA

DeutscheNBGRaiffeisen

Unicredit

BNP Paribas

Credit Agricole

Source: Authors’ estimates based on BankFocus and Central Bank data as of December 2009.Note: BBVA = Banco Bilbao Vizcaya Argentaria, S.A.; NBG = National Bank of Greece.1 Banking groups’ share of outside EU activities through deposit-taking affiliates (see Appendix 6.2).

Figure 6.2 Banks’ Share of Net Income and Capital Outside the EU1

15 The outside EU operations covered by the study only include the opera-tions consolidated on deposit-taking banks, while stand-alone subsid-iaries that perform nonbanking activities (such as insurance companies) or direct cross-border lending (from a parent bank directly to a foreign firm), which are more difficult to ring-fence, are not covered. This partly explains why some banks have a low share of banking assets and profits outside the EU, besides the fact that a large portion of the large EU banks focuses on doing business within the EU only. To refer to a struc-tural level of income and reduce the impact of income volatility, we use five-year 2006–10 averages of net income.

16 For Deutsche Bank, the balance sheet of Deutsche Bank Trust Com-pany Americas has been taken into account, while Deutsche Bank Se-curities, a nonbank, has been excluded—in line with the general approach to include deposit-taking banks only.



ring-fencing context.18 Under full ring-fencing, and assum-ing that European parent banks cannot sell their outside-EU subsidiaries (either because of a lack of supervisor approval or the impossibility of selling without incurring substantial losses) the adjustment could be up to 3 percent of their Core Tier 1 capitalization in Europe (Table 6.1). Banks with impor-tant income and capital outside the EU, such as Santander, NBG, and BBVA, are the most affected in the full ring-fencing context. Santander, Raiffeisen, and Deutsche Bank are the three banks with the highest marginal impact of full ring- fencing, bypassing other banks that are also heavily affected by excess capital being locked.19

The size of the necessary capital adjustments under both partial and full ring-fencing is closely linked to the share of

The partial ring-fencing adjustment deducts, from the EBA profit projections, the share of net income generated outside the EU. The results indicate that the EBA baseline scenario based on consolidated balance sheets overestimates banks’ Core Tier 1 capital ratios by 0 to 0.7 percentage points com-pared with a scenario based on partial ring-fencing (Table 6.1).17 BBVA, NBG, and Santander, all of which have sizable profit buffers partly earned outside the EU, exhibit the largest adjustments to their Core Tier 1 capital ratios. Other banks with large shares of net income from outside the EU, such as HSBC and Unicredit, have relatively lower adjustments in their Tier 1 capital ratios due to the fact that their profit projec-tions under stress are smaller relative to their capital buffers.

The full ring-fencing calculations assume that both the in-come and the excess capital buffers of non-European affili-ates cannot be reallocated to Europe. As expected, the impact for the European part of the international banking group tends to be substantially larger than in the partial

17 Long-term (five-year) profit averages, as often used in stress tests, have been used to avoid drawing conclusions based on overly favorable or unfavorable figures. Moreover, since EBA-reported profit projections are only available on the aggregate, calculations are conservative since the EBA-projected aggregate profits were adjusted as a function of the calculated bank profit share. In the current context, where the crisis center is in Europe, banks’ projections of profits within Europe could be negative in several cases.

18 Theoretically, the adjustment under the full ring-fencing context could imply, if a larger share of bank capital buffers are within EU than out-side, that the European part of an international banking group would end up with a higher Core Tier 1 capital ratio in the full ring-fencing calculations than in no ring-fencing.

19 The difference in the order of banks most affected in the full ring- fencing case versus the partial ring-fencing case captures the full ring-fencing assumption that capital buffers cannot be reallocated. The ring-fencing ad-justments in the EBA adverse scenario projections are not much different in size from those calculated with the EBA baseline scenario figures, but the final adjusted Core Tier 1 capital levels in the adverse scenario are lower since both the ring-fencing adjustments and lower EBA projected profits interact together.

TABLE 6.1

Partial and Full Ring-Fencing Adjustments1

Bank Name Partial Ring-Fencing Adjustment on Tier 1

Capital Ratio (Percent)

Full Ring-Fencing Adjustment on Tier 1


Banco Santander S.A. 0.4 3.1National Bank of Greece 0.5 2.6Banco Bilbao Vizcaya Argentaria S.A. 0.7 1.9Deutsche Bank AG 0.1 1.6Raiffeisen Bank International 0.2 1.4HSBC Holdings PLC 0.3 1.2Barclays PLC 0.1 1.1Dexia 0.1 1.0ING Bank NV 0.3 1.0Unicredit S.p.A. 0.1 0.9Societe Generale 0.1 0.6Rabobank Nederland 0.1 0.5BNP Paribas 0.1 0.5Erste Bank Group 0.1 0.3Swedbank AB 0.0 0.3Royal Bank of Scotland Group 0.0 0.2Nordea Bank AB 0.0 0.1EFG Eurobank Ergasias S.A. 0.0 0.1Alpha Bank 0.0 0.1Credit Agricole 0.0 0.1Commerzbank AG 0.1 0.1Intesa Sanpaolo S.p.A. 0.0 0.1

Sources: Authors’ estimates based on European Banking Authority (EBA); BankFocus; and central bank data.1EBA baseline scenario overestimation of banks’ Core Tier 1 capital ratios in 2012, under the respective ring-fencing scenarios. Only banks with more than EUR 100 billion and 2 percent or more net income share from outside the EU are reported. Remaining banks suffer very small impact. These figures do not correspond with current situations of banks since they are based on 2010 data included in the June 2011 EBA stress test.



to be the least important factor, with correlation coefficients below 0.2 and statistically not different from zero.

Ring-Fencing by Specific Countries or for Specific Subsidiaries

The previous simulation was done by assuming that all non-EU supervisors ring-fence foreign European bank subsidiar-ies. This subsection focuses on evaluating the potential impact of ring-fencing (1) for countries for which there is empirical evidence in the literature; and (2) by assuming ring-fencing in the case of large subsidiaries, which is relevant from a financial stability point of view, whereby the likelihood of ring-fencing by host country supervisors will increase.

banks’ profits, capital, and assets outside Europe. The corre-lation coefficients of these three geographical variables and ring-fencing capital adjustments are high for both capital and income (around 0.8 or more as shown in Table 6.2), and statistically significant. As expected, given the constraints put on capital intragroup movements under full ring- fencing, the correlation between full ring-fencing adjust-ments and the share of capital outside Europe is higher (about 0.92) than under partial ring-fencing and statistically significant. By contrast, the complexity of a banking group—in terms of both the total number of subsidiaries, and also the number of subsidiaries outside Europe—is not highly correlated with the ring-fencing adjustment levels. The size of the banking group—in terms of assets—is found

TABLE 6.2

Correlation Coefficients(Among all banks with a positive capital adjustment need)Variable Partial Ring-Fencing

Adjustment on Tier 1 Capital Ratio


Capital Ratio

Share of Assets Outside Europe 0.82*** 0.77***Share of Capital Outside Europe 0.82*** 0.92***Share of Net Income Outside Europe 0.78*** 0.77***Number of Foreign Bank Subsidiaries in Group 0.23 0.29*Number of Foreign Bank Subsidiaries Outside

Europe0.39** 0.37**

Total Assets of Banking Group 0.10 0.17

Sources: Authors’ estimates based on European Banking Authority; BankFocus; and central bank data.Note: Correlation coefficients among all banks without a capital adjustment need. *** = significance at 1 percent level; ** = significance at 5 percent level; * = significance at 10 percent level.

TABLE 6.3

Partial and Full Ring-Fencing Adjustments for Countries with Anecdotal Evidence1





National Bank of Greece 0.4 2.4KBC Bank 0.9 1.2Unicredit 0.2 1.2Dexia 0.1 1.0Oesterreichische Volksbank AG 0.1 0.9Raiffeisen Bank International 0.2 0.5Commerzbank 0.3 0.4ING Bank 0.2 0.4Societe Generale 0.2 0.4Erste Bank 0.5 0.4EFG Eurobank 0.0 0.1Rabobank 0.0 0.1Banco Comercial Portugues 0.1 0.1BNP Paribas 0.0 0.1Deutsche Bank 0.0 0.1Piraeus Bank 0.0 0.1Nordea 0.0 0.1

Sources: Authors’ estimates based on European Banking Authority (EBA); Bankscope; and central bank data.1EBA baseline scenario overestimation of banks’ Core Tier 1 capital ratios in 2012, under the respective ring-fencing scenarios. Only banks with adjustments larger than 0.1 percent are reported. These figures do not correspond with current situations of banks since they are based on 2010 data included in the June 2011 EBA stress test.



about 56 deposit-taking subsidiaries, shows that ring-fencing can not only be an important risk for banking groups’ sol-vency but may also erode banks’ diversification strategies21 under stress (which was, for example, a stabilizing factor in some peripheral European countries during recent years but not in others) or business strategies to cope with relatively low structural income levels in their home markets (such as in Germany, France, Switzerland, and Japan, for example22). This is especially evident from the relatively high impact of partial ring-fencing, which is affecting banks with important subsidiaries in those five countries such as KBC and Erste Bank.

The maximum calculated impact of both partial and full ring-fencing is also similar in the third case study, where ring-fencing is applied only to those subsidiaries that exhibit 5 percent or more of the total banking assets in their

Table 6.3 provides the outcome of the analysis of partial and full ring-fencing for five countries in Central and Eastern Europe with anecdotal evidence of ring-fencing in Albania, Croatia, the Czech Republic, Poland, and Turkey—all coun-tries with a significant portion of foreign-owned banks.20 De-spite assuming ring-fencing by only a few host supervisors, the results indicate a broadly similar important impact on some banking groups, both in terms of partial (up to 0.9 per-cent) and full ring-fencing (up to 2.4 percent). These banking groups are smaller in size relative to the entire sample, but the ring-fencing assumptions capture large and profitable bank-ing subsidiaries within each group (that is, the sample corre-lation between banking group size and the share of capital and net income is about zero). This scenario, which covers

20 For three of the five countries (Albania, Croatia, Czech Republic), the evolution of capital ratios relative to the country average from 2006 to 2010 shows an upward trend for the foreign subsidiaries. This suggests that foreign affiliates were either treated differently by supervisors or decided to increase their capital levels relative to the country average for strategic reasons. Other countries for which this indirect evidence of ring-fencing was present were Brazil, Canada, Georgia, Russia, Serbia, and the United States. These countries were not included because the number of observations is very low for some and because they were al-ready part of the previous non-EU simulation.

21 Diversification benefits of business strategies are recognized by the reg-ulatory framework, for example, under Pillar 1 (intra-risk diversifica-tion for market risk) and Pillar 2 (for example, for the European Union, inter-risk diversification (that is, between risks) and intragroup diversi-fication (that is, diversification across countries). See EBA 2010.

22 See page 54 of the Bank for International Settlements 83rd Annual Re-port (BIS 2013), for example.

TABLE 6.4

Partial and Full Ring-Fencing Adjustments for Countries for Large Subsidiaries1





National Bank of Greece 0.5 2.9Banco Santander S.A. 0.4 2.8Erste Bank Group (EBG) 0.9 2.1Banco Bilbao Vizcaya Argentaria S.A. (BBVA) 0.7 1.8Unicredit S.p.A. 0.2 1.4Swedbank AB (publ) 0.1 1.2Skandinaviska Enskilda Banken AB (publ) (SEB) 0.1 1.1Raiffeisen Bank International (RBI) 0.5 1.1Dexia 0.1 1.0KBC Bank 0.4 1.0Societe Generale 0.3 0.8Piraeus Bank Group 0.0 0.5Commerzbank AG 0.3 0.4HSBC Holdings PLC 0.1 0.4Oesterreichische Volksbank AG 0.1 0.3Iintesa Sanpaolo S.p.A. 0.1 0.2EFG Eurobank Ergasias S.A. 0.0 0.2ING Bank NV 0.1 0.2Bayerische Landesbank 0.0 0.1Banco Comercial Portugues, S.A. (BCP) 0.1 0.1Royal Bank of Scotland Group PLC 0.0 0.1BNP Paribas 0.0 0.1

Sources: Authors’ estimates, based on European Banking Authority (EBA); Bankscope; and central bank data.Note: publ = publicly traded.1EBA baseline scenario overestimation of banks’ Core Tier 1 capital ratios in 2012, under the respective ring-fencing scenarios. Only banks with adjustments larger than 0.1 percent are reported. These figures do not correspond with current situations of banks since they are based on 2010 data included in the June 2011 EBA stress test.



for and the incidence of ring-fencing by the host country authorities. Without a credible cross-border resolution framework aligning incentives, host country regulators may ring-fence due to macro-financial stability considerations, such as the need to protect the domestic banking system from negative spillovers from the rest of the group, or more generally, to increase reserves for their domestic banking sys-tem at times when the impending output collapses and bank losses associated with a crisis abroad are uncertain.

Second, in the current context of non-fully-fledged reso-lution and burden-sharing mechanisms (even among EU members), minimum regulatory capital levels for cross- border banking groups would have to take into account the potential risks of ring-fencing, especially during crisis times. More work on estimating different ring-fencing scenarios across international banks is needed in order to assess the potential additional capital buffer needs at the group levels, which could lower the likelihood of ring-fencing. These larger capital buffers—together with sound liquidity lev-els—would not only provide capital buffers in the potential presence of ring-fencing, but they could also reassure host country regulators that the banking group has enough capi-tal buffers to withstand ring-fencing by other host regula-tors, thereby avoiding preemptive actions. In other words, host regulators in a country would have fewer incentives to ring-fence well-capitalized foreign subsidiaries just in antici-pation of what would be the impact on the banking group when other host regulators ring-fence the banking group subsidiaries under their supervision.

Third, the analysis also highlights that it is not just the level of a banking group’s capital buffers that is relevant but also that the geographical location of those buffers within the banking group (which are not necessarily highly corre-lated with the group size or its complexity) matters. The need for higher capital buffers for cross-border banking groups could be larger if some recent reform proposals, ratio-nal from individual country perspectives (for example, sepa-rating UK retail business from the rest of the banking group and increasing capital buffers on those operations), trigger new higher levels of ring-fencing—even among Organisa-tion for Economic Co-operation and Development coun-tries. Moreover, even without a crisis, if the geographical distribution of banking groups’ capital buffers changes enough (due to regulatory reforms in parent banking group countries), it is not unthinkable that (some) host country regulators might also increase subsidiaries’ capital require-ments—even if this involves covering both domestic and foreign banks—in order to offset significant changes in the geographical distribution of capital buffers. These elements highlight that further international cooperation is essential in order to avoid undesired outcomes.

respective host country outside the EU or in Emerging EU countries (Table 6.4). This case covers about 100 deposit-tak-ing subsidiaries, which are not only important for the host country but also for the analyzed banking groups as reflected in the increase in average share of capital and net income in those important subsidiaries (Appendix 6.2). The inclusion of both important affiliates operating in both non-EU countries and Emerging EU countries brings together the potential vulnerabilities of two types of banking groups: those banking groups that have very important subsidiaries outside Europe relative to the host market (for example, Santander, BBVA, and NBG) and those that are more exposed to Eastern Eu-rope (for example, Unicredit, Erste bank, and KBC). Com-pared with the non-EU scenario (Table 6.1), the current scenario highlights that some banks with subsidiaries outside the EU, which might be large relative to the group in the ag-gregate, but not necessarily relative to the host market (for example, Deutsche Bank, HSBC, and Barclays), will likely be less vulnerable to worldwide non-EU ring-fencing if it is lim-ited to only systemic affiliates.

In summary, the case studies highlight the potential im-portant impact of ring-fencing by host country supervisors across different scenarios. The affected banking groups would be a function not only of which host supervisors ring-fence their foreign affiliates, but also the way that they do it (targeted or broad ring-fencing). In general, the key determi-nant seems to be the volume of cross-border activity (in terms of share of capital and net income) through foreign affiliates potentially exposed to be ring-fenced, rather than banking group size.

4. CONCLUSIONThe analysis presented in this chapter is an example of how the proposed group-structure approach extends best practice stress tests if one seeks to gain a more comprehensive under-standing of risks faced by international banking groups. As shown, provided that granular data are made available (best) or proxied (second best), the outcome of group-structure stress tests taking into account different degrees of ring- fencing can differ considerably from traditional consolidated top-down stress tests. Moreover, the need for explicitly tak-ing into account the business structure of international banks when running a group-structure approach provides a better understanding of the potential vulnerabilities of the single (large) banks within the group.

What are the policy implications of this analysis? First, the establishment of a credible framework for the resolution of cross-border banking groups would help to avoid unilat-eral and likely more costly solutions (in terms of capital re-quirements). Such frameworks could reduce the incentives


Appendix 6.1.Background Information on European

Banking Authority Stress Tests

Appendix Table 6.1.1 provides an overview of recent European and US stress tests. As shown, both exercises covered more than half of the respective assets of the banking systems. Stress was projected for two to three years, based on macroeconomic sce-narios broadly replicating the global financial crisis. The pass rate was higher than the regulatory minimum, and gradually in-creasing over time, reflecting the purpose of the tests to calm down capital markets.

In terms of the macroeconomic scenarios, the European Banking Authority (EBA) came up with a global scenario, which was then broken down into country-specific trajectories of GDP growth, inflation, and unemployment rates for each of the 27 EU member countries, and scenarios for major non-EU countries. Trajectories were also provided for government bond yields, stock prices, and house prices for all EU members.23 Banks then translated these scenarios into changes of key solvency param-eters, based on the guidance provided by EBA, and subject to discussions with the supervisors.

23 See http://www.eba.europa.eu/-/the-eba-publishes-details-of-its-stress-test-scenarios-and-methodology for more information.


Overview of Recent European and US Stress TestsTest US 2009 (SCAP) EU 2010 EU 2011 US 2012 (CCAR)

Number of banks 19 (>60 percent of assets) 91 (65 percent of assets) 90 18Projection

horizonTwo years (until end of 2010) Two years (until end of 2011) Two years (until end of 2012) Three years (until end of

2014)Pass rate 4 percent Common Equity

Tier 16 percent Tier 1 5 percent Core Tier 1 5 percent Common Equity

Tier 1Scenario

(adverse vs. baseline)

Cumulative drop of GDP by 3 percentage points




Sources: EBA 2011; and Federal Reserve Board 2012.Note: CCAR = Comprehensive Capital Analysis and Review; SCAP = Supervisory Capital Assessment Program.


http://www.eba.europa.eu/-/the-eba-publishes-details-of-its-stress-test-scenarios-and-methodology

Appendix 6.2.Mapping Bank Groups

The case studies are conducted using publicly available data on bank affiliates (the second-best solution in Section 2 given the lack of access to supervisory data. The analysis covers 51 banking groups, with large cross-border activity, from the 90 banking groups included in the June 2011 European Banking Authority stress test (see summary statistics in Appendix Table 6.2.1). Banking groups included covered the main EU banking systems (see Appendix Tables 6.2.2 and 6.2.3), leaving aside smaller domestic banking groups (for example, from Cyprus, Hungary, Malta, Poland, Slovenia, and so on) and 24 Spanish regional banks that went through successive mergers and restructuring processes even during 2009–10.

The geographical presence of each banking group was built by mapping, as of December 2009, the authorized deposit-taking institutions of several advanced and emerging countries (see list of countries below). Once the affiliates of each European Banking Authority banking group were identified, balance sheet data were downloaded from Bankscope, and, if data were not available there, then bank regulators’ websites were used as a secondary source (this applied to about 15 percent of the total affiliates). In total, the analysis identified about 490 affiliates, of which about 170 were outside the EU. The final coverage for foreign affiliate data was almost complete in most non-EU countries (with the exception of Algeria and Jordan) and emerging EU countries, but comprehensive balance sheet information on affiliates could not be collected in several advanced EU countries (France, Luxem-bourg, Netherlands). This was especially the case for branches. However, this is unlikely to trigger large biases because (1) the simulations are not focusing on the euro area, in which banks have many branches (rather than subsidiaries); and (2) it is more difficult for branches to be ring-fenced by host countries since their supervision is carried out by the home supervisor. About 10 branches in several emerging countries—where they are required to hold capital as subsidiaries—were included in the analysis.

EU+ Countries: Norway, Switzerland, and the 27 current EU members (Austria, Belgium, Bulgaria,* Cyprus, Czech Re-public,* Denmark, Estonia,* Finland, France, Germany, Greece, Hungary,* Ireland, Italy, Latvia,* Lithuania,* Luxembourg, Malta, Netherlands, Poland,* Portugal, Romania,* Slovakia,* Slovenia,* Spain, Sweden, and United Kingdom). Countries with “*” were included among the emerging EU countries subgroup.

Non-EU Countries: Albania, Algeria, Argentina, Armenia, Australia, Barbados, Belarus, Bolivia, Bosnia, Brazil, Canada, Chile, China, Colombia, Costa Rica, Croatia, Dominican Republic, Egypt, El Salvador, Georgia, Guatemala, Honduras, India, Indonesia, Jamaica, Japan, Kazakhstan, Korea, Malaysia, Mexico, Morocco, New Zealand, Nicaragua, Pakistan, Panama, Paraguay, Peru, Philippines, Russia, Serbia, Thailand, Trinidad and Tobago, Tunisia, Turkey, Ukraine, Uruguay, United States, and Venezuela.


Summary of Banking Groups’ Geographical Presence (51 Banking Groups)

Average Median Standard Deviation Minimum Maximum

Total group assets (USD billions)1 735.9 343.7 785.9 43.3 2964.3Number of affiliates by group 10 7 9 0 38Number of affiliates outside EU2 3 1 5 0 29Percent of assets outside EU2 3.5 0.6 6.5 0.0 25.2Percent of capital outside EU2 6.5 1.6 10.3 0.0 43.0Percent of net income outside EU2,3 5.3 1.0 8.2 0.0 27.4Number of affiliates in countries with anecdotal ring-fencing4 1 0 1 0 6Percent of assets in countries with anecdotal ring-fencing4 2.4 0.1 5.1 0.0 21.3Percent of capital in countries with anecdotal ring-fencing4 3.6 0.1 6.8 0.0 25.2Percent of net income in countries with anecdotal ring-fencing3,4 6.4 0.1 15.0 0.0 78.3Number of nonsystemic subsidiaries5 2 1 3 0 13Percent of assets in nonsystemic subsidiaries5 5.0 0.3 8.0 0.0 39.0Percent of capital in nonsystemic subsidiaries5 7.3 0.6 10.9 0.0 43.0Percent of net income in nonsystemic subsidiaries3,5 9.8 0.6 17.2 0.0 78.3

Source: Authors’ estimates based on Bankscope and central bank data.1Bank Group Assets as of December 2009.2Number or share of total asset/capital/net income located in deposit-taking affiliates operating in selected Organisation for Economic Co-operation and Development and emerging economies.3Based on 2006–10 average.4Number or share of total asset/capital/net income located in deposit-taking affiliates in Albania, Croatia, the Czech Republic, Poland, or Turkey.5Number or share of total asset/capital/net income located in systemic deposit-taking affiliates in Emerging Europe and non-EU countries. Systemic subsidiaries in the sense that their assets are at least larger than 5 percent of host country market share.




Banking Groups’ Geographical PresenceBank Name Parent Group

CountryTotal Group Assets (USD

billions)1

Percent of Assets outside

EU2

Percent of Capital

outside EU2

Percent of Net Income

outside EU2,3

Erste Bank Group Austria 290.6 4.0 6.3 6.4Raiffeisen Bank International Austria 213.5 19.5 29.8 18.8Oesterreichische Volksbank AG Austria 69.5 4.5 17.9 4.5Dexia Belgium 832.1 2.1 8.8 20.3KBC Bank Belgium 426.1 1.0 2.3 0.0Deutsche Bank AG Germany 2161.8 3.5 20.0 8.5Commerzbank AG Germany 1216.0 0.3 0.7 19.2Landesbank Baden-Württemberg Germany 593.1 0.0 0.0 0.0DZ Bank AG Germany 559.7 0.0 0.0 0.0Bayerische Landesbank Germany 488.1 0.4 1.5 0.4Norddeutsche Landesbank-GZ Germany 343.7 0.0 0.0 0.0Hypo Real Estate Holding AG Germany 518.1 0.0 0.0 0.0WestLB AG Germany 349.1 0.5 2.2 0.5HSH Nordbank AG Germany 251.4 0.0 0.0 0.0Landesbank Berlin AG Germany 205.1 0.0 0.0 0.0DekaBank Deutsche Girozentrale Germany 192.0 0.0 0.0 0.0WGZ Bank AG Germany 137.8 0.0 0.0 0.0Danske Bank Denmark 597.0 0.0 0.1 0.1Jyske Bank Denmark 43.3 0.0 0.0 0.0Banco Santander S.A. Spain 1616.9 22.9 43.0 27.4Banco Bilbao Vizcaya Argentaria S.A. Spain 734.0 25.2 32.1 23.5Banco Popular Espanol S.A. Spain 186.3 0.0 0.0 0.0BNP Paribas France 2964.3 3.8 8.3 5.6Credit Agricole France 2440.0 0.6 1.6 1.9BPCE France 1482.1 0.0 0.2 1.0Societe Generale France 1474.7 3.3 9.4 6.1Royal Bank of Scotland Group PLC United Kingdom 2747.4 15.1 17.7 15.1HSBC Holdings PLC United Kingdom 2364.5 22.3 28.4 26.5Barclays PLC United Kingdom 2233.2 6.7 15.9 6.2Lloyds Banking Group PLC United Kingdom 1663.6 0.0 0.0 0.0EFG Eurobank Ergasias S.A. Greece 121.4 2.2 3.8 2.2National Bank of Greece Greece 163.4 13.8 27.0 21.1Alpha Bank Greece 100.3 1.4 2.5 2.1Piraeus Bank Group Greece 78.2 4.3 4.6 3.4Agricultural Bank of Greece S.A. Greece 47.3 0.0 0.0 0.0Allied Irish Banks PLC Ireland 251.1 0.0 0.0 0.0Intesa Sanpaolo S.p.A. Italy 900.1 1.1 1.7 2.0Unicredit S.p.A. Italy 1338.0 7.4 16.6 21.8Banca Monte dei Paschi di Siena S.p.A. Italy 323.9 0.0 0.0 0.0Banco Popolare - S.C. Italy 195.5 0.0 0.0 0.0Unione di Banche Italiane SCPA Italy 176.2 0.0 0.0 0.0ING Bank NV Netherlands 1270.8 7.6 12.6 12.2Rabobank Nederland Netherlands 875.4 1.7 5.1 4.5SNS Bank NV Netherlands 115.7 0.0 0.0 0.0DnB NOR Bank ASA Norway 315.5 0.1 1.2 0.0Banco Comercial Portugues (BCP) Portugal 137.6 0.6 0.9 0.3Espirito Santo Financial Group Portugal 118.6 2.0 2.3 4.1Nordea Bank AB (publ) Sweden 731.2 0.7 1.9 1.0Skandinaviska Enskilda Banken AB (publ) Sweden 324.3 0.0 0.1 0.1Svenska Handelsbanken AB (publ) Sweden 298.3 0.0 0.0 0.0Swedbank AB (publ) Sweden 252.2 1.9 4.4 2.0

Source: Authors’ estimates based on Bankscope and central bank data.Note: publ = publicly traded.1Bank Group Assets as of December 2009.2Share of total asset/capital/net income located in deposit-taking affiliates operating in selected Organisation for Economic Co-operation and Develop-ment countries and in emerging countries.3Based on 2006–10 average.


Eugenio Cerutti and C

hristian Schmieder

115



CountryTotal Group

Assets (USD billions)1

Percent of Assets in

Countries with Anecdotal

Ring-Fencing2

Percent of Capital in


Ring-Fencing2

Percent of Net Income in


Ring-Fencing2,3


Nonsystemic Subsidiaries4




Nonsystemic Subsidiaries3,4

Erste Bank Group Austria 290.6 21.3 19.2 41.2 39.0 43.0 76.9Raiffeisen Bank International Austria 213.5 16.4 18.8 20.3 27.6 31.2 38.9Oesterreichische Volksbank AG Austria 69.5 5.9 19.7 5.9 11.7 15.2 11.7Dexia Belgium 832.1 2.1 8.8 20.3 2.5 9.5 18.9KBC Bank Belgium 426.1 18.4 19.3 52.7 17.7 20.6 26.4Deutsche Bank AG Germany 2161.8 0.4 1.4 1.1 0.1 0.1 0.1Commerzbank AG Germany 1216.0 2.3 3.7 78.3 2.3 3.7 78.3Landesbank Baden-Württemberg Germany 593.1 0.3 0.3 0.3 0.0 0.0 0.0DZ Bank AG Germany 559.7 0.1 0.5 0.6 0.0 0.0 0.0Bayerische Landesbank Germany 488.1 0.0 0.0 0.0 3.4 4.6 4.3Norddeutsche Landesbank GZ Germany 343.7 0.0 0.0 0.0 0.0 0.0 0.0Hypo Real Estate Holding AG Germany 518.1 0.0 0.0 0.0 0.0 0.0 0.0WestLB AG Germany 349.1 0.0 0.0 0.0 0.0 0.0 0.0HSH Nordbank AG Germany 251.4 0.0 0.0 0.0 0.0 0.0 0.0Landesbank Berlin AG Germany 205.1 0.0 0.0 0.0 0.0 0.0 0.0DekaBank Deutsche Girozentrale Germany 192.0 0.0 0.0 0.0 0.0 0.0 0.0WGZ Bank AG Germany 137.8 0.0 0.0 0.0 0.0 0.0 0.0Danske Bank Denmark 597.0 0.1 0.3 0.3 0.0 0.0 0.0Jyske Bank Denmark 43.3 0.0 0.0 0.0 0.0 0.0 0.0Banco Santander S.A. Spain 1616.9 0.0 0.1 0.0 18.2 37.6 27.3Banco Bilbao Vizcaya Argentaria S.A. Spain 734.0 0.0 0.0 0.0 16.1 23.5 23.3Banco Popular Espanol S.A. Spain 186.3 0.0 0.0 0.0 0.0 0.0 0.0BNP Paribas France 2964.3 0.7 1.5 1.2 0.4 0.9 1.0Credit Agricole France 2440.0 0.1 0.3 0.1 0.3 0.6 0.6BPCE France 1482.1 0.0 0.0 0.0 0.0 0.0 0.0Societe Generale France 1474.7 3.5 6.3 13.4 5.3 10.9 20.8Royal Bank of Scotland Group PLC United Kingdom 2747.4 0.1 0.1 0.1 10.6 11.3 18.7HSBC Holdings PLC United Kingdom 2364.5 0.0 0.1 0.0 5.9 8.4 7.0Barclays PLC United Kingdom 2233.2 0.0 0.0 0.0 0.0 0.0 0.0Lloyds Banking Group PLC United Kingdom 1663.6 0.0 0.0 0.0 0.0 0.0 0.0EFG Eurobank Ergasias S.A. Greece 121.4 2.2 3.8 2.2 8.1 10.4 9.1National Bank of Greece Greece 163.4 13.0 25.2 20.3 16.1 30.3 22.9Alpha Bank Greece 100.3 0.0 0.0 0.0 7.2 6.1 4.6Piraeus Bank Group Greece 78.2 1.1 2.0 2.5 4.5 9.3 8.9Agricultural Bank of Greece S.A. Greece 47.3 0.0 0.0 0.0 0.0 0.0 0.0Allied Irish Banks PLC Ireland 251.1 7.6 10.7 7.6 7.6 10.7 7.6Intesa Sanpaolo S.p.A. Italy 900.1 0.1 0.2 0.3 4.4 5.8 8.5Unicredit S.p.A. Italy 1338.0 9.6 20.5 29.7 12.8 25.3 35.7Banca Monte dei Paschi di Siena S.p.A. Italy 323.9 0.0 0.0 0.0 0.0 0.0 0.0Banco Popolare S.C. Italy 195.5 0.1 0.4 0.1 0.0 0.0 0.0Unione di Banche Italiane SCPA Italy 176.2 0.0 0.0 0.0 0.0 0.0 0.0ING Bank NV Netherlands 1270.8 2.5 4.3 7.6 1.7 2.3 4.8

(continued)



APPENDIX TABLE 6.2.3 (continued)


CountryTotal Group

Assets (USD billions)1



Ring-Fencing2



Ring-Fencing2



Ring-Fencing2,3






Nonsystemic Subsidiaries3,4

Rabobank Nederland Netherlands 875.4 1.2 1.9 1.7 0.2 0.3 0.3SNS Bank NV Netherlands 115.7 0.0 0.0 0.0 0.0 0.0 0.0DnB NOR Bank ASA Norway 315.5 0.9 1.3 0.2 1.2 1.2 1.2

Banco Comercial Portugues (BCP) Portugal 137.6 12.0 12.2 18.7 11.4 11.3 18.5Espirito Santo Financial Group Portugal 118.6 0.0 0.0 0.0 0.0 0.0 0.0Nordea Bank AB (publ) Sweden 731.2 1.0 1.5 0.9 0.0 0.0 0.0Skandinaviska Enskilda Banken AB

(publ)Sweden 324.3 0.0 0.0 0.0 7.4 14.7 10.9

Svenska Handelsbanken AB (publ) Sweden 298.3 0.0 0.0 0.0 0.0 0.0 0.0Swedbank AB (publ) Sweden 252.2 0.0 0.0 0.0 12.5 21.6 12.5

Source: Authors’ estimates based on Bankscope and central bank data.Note: publ = publicly traded.1Bank Group Assets as of December 2009.2Share of total asset/capital/net income located in deposit-taking affiliates in Albania, Croatia, the Czech Republic, Poland, or Turkey.3Based on 2006–10 average.4Share of total asset/capital/net income located in systemic deposit-taking affiliates in Emerging Europe and non-EU countries. Systemic subsidiaries in the sense that their assets are at least larger than 5 per-cent of host country market share.



European Bank for Reconstruction and Development (EBRD). 2013. “Unilateral Measures to Safeguard National Financial Stability.” Transition Report 2012, Chapter 3. London: Euro-pean Bank for Reconstruction and Development. http://2012 .tr-ebrd.com/chapter-3/box-3-4-unilatera l-measures-to -safeguard-national-financial-stability.html.

Federal Reserve Board. 2012. Comprehensive Capital Analysis and Review 2012: Methodology for Stress Scenario Projection. Wash-ington, DC: Federal Reserve Board. https://www.federalreserve .gov/supervisionreg/ccar-2012.htm.

Financial Stability Board. 2012. Update of Group of Global Systemi-cally Important Banks (G-SIBs). Basel: Financial Stability Board. http://www.fsb.org/2012/11/r_121031ac/.

Foglia, Antonella. 2009. “Stress Testing Credit Risk: A Survey of Authorities’ Approaches.” International Journal of Central Bank-ing 5 (3): 9–45.

Freixas, Xavier. 2003. “Financial Supervision in Europe.” In Crisis Management in Europe, edited by J. Kremers, D. Schoenmaker, and P. Wierts. Cheltenham: Edward Elgar.

Hardy, Daniel, and Maria J. Nieto. 2011. “Cross-border Coordina-tion of Prudential Supervision and Deposit Guarantees.” Jour-nal of Financial Stability 7 (3): 155–164.

International Monetary Fund (IMF). 2010. “Resolution of Cross-Border Banks—A Proposed Framework for Enhanced Coordi-nation.” IMF Policy Paper, Washington, DC. https://www.imf .org/en/Publicat ions/Policy-Papers/Issues/2016/12/31 /Resolution-of-Cross-Border-Banks-A-Proposed-Framework -for-Enhanced-Coordination-PP4462.

———. 2012a. “Czech Republic—Technical Note on Stress Testing the Banking Sector.” IMF Country Report 12/174, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /Czech-Republic-Technical-Note-on-Stress-Testing-the-Banking -Sector-26067.

———. 2012b. “Spain: Financial System Stability Assessment.” IMF Country Report 12/137, Washington, DC. https://www.imf.org /en/Publications/CR/Issues/2016/12/31/Spain-Financial-System -Stability-Assessment-25977.

Krimminger, Michael H. 2008. “The Resolution of Cross-Border Banks: Issues for Deposit Insurers and Proposal for Coopera-tion.” Journal of Financial Stability 4 (4): 376–390.

Ong, Li Lian, and Martin Čihák. 2010. “Of Runes and Sagas: Perspec-tive on Liquidity Stress Testing Using an Iceland Example.” IMF Working Paper 10/156, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2016 /12/31/Of-Runes-and-Sagas-Perspectives-on-Liquidity-Stress -Testing-Using-an-Iceland-Example-24019.

Schoenmaker, Dirk, and Arjen Siegmann. 2014. “Can European Bank Bailouts Work?” Journal of Banking and Finance 48: 334–349.

Schmieder, Christian, Heiko Hesse, Benjamin Neudorfer, Claus Puhr, and Stefan Schmitz. 2012. “Next Generation System-Wide Liquidity Stress Testing.” IMF Working Pape 03/12, In-ternational Monetary Fund, Washington, DC. https://www .imf.org/external/pubs/cat/longres.aspx?sk=25509.0.

van Lelyveld, Iman, and Marco Spaltro. 2011. “Coordinating Bank Failure Costs and Financial Stability.” DNB Working Paper 306, De Nederlandsche Bank, Amsterdam. https://www.dnb .nl/en/news/dnb-publications/dnb-working-papers-series/dnb -working-papers/working-papaers-2011/dnb256308.jsp.

REFERENCESAvgouleas, Emilios, Charles Goodhart, and Dirk Schoenmaker.

2013. “Bank as a Catalyst for Global Financial Reform.” Jour-nal of Financial Stability 9 (2): 210–18.

Bank for International Settlements (BIS). 2013. 83rd Annual Re-port 2012/13. Basel: Bank for International Settlements. https://www.bis.org/publ/arpdf/ar2013e.htm.

Basel Committee on Banking Supervision. 2011. Global Systemi-cally Important Banks: Assessment Methodology and the Additional Loss Absorbency Requirement. Basel: Bank for International Set-tlements. https://www.bis.org/publ/bcbs207.htm (updated in December 2018, https://www.bis.org/bcbs/gsib/).

Borio, Claudio, Drehmann, Mathias, and Kostas Tsatsaronis. 2014. “Stress-testing Macro Stress Testing: Does It Live up to Expectations?” Journal of Financial Stability 12 (1): 3–15.

Cerutti, Eugenio, Stijn Claessens, and Patrick McGuire. 2014. “Systemic Risks in Global Banking: What Available Data Can Tell and What More Data are Needed?” In Risk Topography: Systemic Risk and Macro Modeling, edited by M. Brunnermeier and A. Krishnamurthy. Chicago: The University of Chicago Press.

Cerutti, Eugenio, Giovanni Dell’Ariccia, and Maria S. Martinez Peria. 2007. “How banks go Abroad? Branches or Subsidiaries.” Journal of Banking and Finance 31 (6): 1669–92.

Cerutti, Eugenio, Anna Ilyina, Yuliya Makarova, and Christian Schmieder. 2010. “Bankers without Borders? Implications of Ring-fencing for European Cross-Border Banks.” IMF Working Paper 10/247, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 /Bankers-Without-Borders-Implications-of-Ring-Fencing -for-European-Cross-Border-Banks-24335.

Cerutti, Eugenio, and Christian Schmieder. 2014. “Ring Fencing and Consolidated Banks’ Stress Tests.” Journal of Financial Sta-bility 11 (C): 1–12

Cetorelli, Nicola, and Linda S. Goldberg. 2012a. “Liquidity Man-agement of US Global Banks: Internal Capital Markets in the Great Recession.” Journal of International Economics 88 (2): 299–311.

———. 2012b. “Banking Globalization and Monetary Transmis-sion.” Journal of Finance 67 (5): 1811–1843.

Dell’Ariccia, Giovanni, and Robert Marquez. 2006. “Competition across Regulators and Credit Market Integration.” Journal of Financial Economics 79 (2): 401–430.

De Haas, Ralph, and Iman van Lelyveld. 2010. “Internal Capital Markets and Lending by Multinational Bank Subsidiaries.” Journal of Financial Intermediation 19(1): 1–25.

Eisenbeis, Robert A., and George G. Kaufman. 2008. “Cross- border Banking and Financial Stability in the EU.” Journal of Financial Stability 4 (3): 168–204.

European Banking Authority (EBA). 2010. Guidelines on the Man-agement of Concentration Risk under the Supervisory Review Pro-cess (GL31). London: European Banking Authority. http://www.eba.europa.eu/regulation-and-policy/supervisory-review /guidelines-on-the-management-of-concentration-risk-under -the-supervisory-review-process.

———. 2011. “2011 EU-Wide Stress Test: Methodological Note.” European Banking Authority, London. https://www.eba.europa .eu/-/the-eba-publishes-details-of-its-stress-test-scenarios-and -methodology.


http://2012.tr-ebrd.com/chapter-3/box-3-4-unilateral-measures-to-safeguard-national-financial-stability.html



https://www.federalreserve.gov/supervisionreg/ccar-2012.htm

https://www.federalreserve.gov/supervisionreg/ccar-2012.htm

http://www.fsb.org/2012/11/r_121031ac/

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Resolution-of-Cross-Border-Banks-A-Proposed-Framework-for-Enhanced-Coordination-PP4462.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Czech-Republic-Technical-Note-on-Stress-Testing-the-Banking-Sector-26067.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Spain-Financial-System-Stability-Assessment-25977






https://www.imf.org/external/pubs/cat/longres.aspx?sk=25509.0.van

https://www.imf.org/external/pubs/cat/longres.aspx?sk=25509.0.van

https://www.dnb.nl/en/news/dnb-publications/dnb-working-papers-series/dnb-working-papers/working-papaers-2011/dnb256308.jsp



https://www.bis.org/publ/arpdf/ar2013e.htm



https://www.bis.org/bcbs/gsib/

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Bankers-Without-Borders-Implications-of-Ring-Fencing-for-European-Cross-Border-Banks-24335



http://www.eba.europa.eu/regulation-and-policy/supervisory-review/guidelines-on-the-management-of-concentration-risk-under-the-supervisory-review-process.




https://www.eba.europa.eu/-/the-eba-publishes-details-of-its-stress-test-scenarios-and-methodology



CHAPTER 7

Rules of Thumb for Bank Solvency Stress Testing

DANIEL C. HARDY • CHRISTIAN SCHMIEDER

Rules of thumb can be useful in undertaking quick, robust, and readily interpretable bank stress tests. Such rules of thumb are proposed for the behavior of banks’ capital ratios and key drivers thereof— primarily credit losses, income, credit growth, and risk weights— in advanced and

emerging market economies, under more or less severe stress conditions. The proposed rules imply disproportionate responses to large shocks, and can be used to approximate the cyclical behavior of capital ratios under various regulatory approaches.

guidelines, particularly to understand worst- case scenarios, and to complement, rather than substitute for, detailed anal-ysis of a country’s banks and its institutional, financial, and conjunctural circumstances.

Rules of thumb would be useful at various stages of stress testing:

• For countries with limited relevant experience and data: One challenge is to design tests for a country with limited relevant experience and data, either because of structural breaks in time series or because the county did not suffer a severe banking crisis in recent decades. Then, on the one hand, rules of thumb derived from experience in various countries with underlying simi-larities can be used to construct and calibrate relevant stress tests for the country concerned. Rules of thumb based on long, worldwide experience provide indica-tors of the great variety of scenarios that countries have suffered. On the other hand, given a scenario, rules of thumb for behavioral relationships can be used to make projections when reliable stress testing methods are unavailable locally. For example, when national authorities or bank management are unable to estimate behavioral relationships robustly based

1. INTRODUCTIONFinancial sector stress testing has become a widespread, im-portant, and prominent activity. Stress testing is used to identify financial sector vulnerabilities,1 to inform policy de-cisions affecting the financial system and individual institu-tions,2 and to guide companies’ own risk management. Since the crisis, its focus has gradually shifted from crisis manage-ment (“crisis stress testing”) (see Ong and Pazarbasioglu 2014) to a more general assessment of the adequacy of bank health. Yet, its practical application is often demanding (for example, compare the length of the underlying documents for the Comprehensive Capital Analysis and Review and Eu-ropean Banking Authority stress tests in 2011 and 2016–17), and there remain questions about the robustness of its results.

Obtaining stress test results and establishing their robust-ness would be facilitated by the availability of “rules of thumb,” that is, rough guides to typical behavioral relation-ships, magnitudes of shocks, and the impact of shocks on banks, based on a wide range of experience. This chapter at-tempts to identify such rules of thumb that apply to bank solvency, that is, capital ratios. The rules of thumb elabo-rated in this chapter are meant to serve mainly as general

The authors would like to thank Martin Čihák, Sam Langfield, Li Lian Ong, and Mario Quagliariello for helpful comments and suggestions. This chapter is based on IMF Working Paper 13/232 (Hardy and Schmieder 2013).1 For example, macro stress tests as part of Financial Sector Assessment Programs (Jobst, Ong, and Schmieder 2013). See Ong 2014 for an overview of

approaches.2 Prominent examples are the US regulatory stress tests in 2009, the Supervisory Capital Assessment Program (Board of Governors of the Federal Reserve

System 2009), and since 2011 on an annual basis (Comprehensive Capital Analysis and Review), and the European Stress Tests conducted in 2011, 2014, and 2016 by the European Banking Authority, and in 2009–10 by the Committee of European Banking Supervisors.


Rules of Thumb for Bank Solvency Stress Testing120

regarding the potential outcome of stress tests and some of the main behavioral relationships underlying them.

Motivated by these considerations, this chapter concen-trates on the formulation of rules of thumb for key factors affecting bank solvency. They comprise credit losses, preim-pairment income, and credit growth during crises and illustrate their use in the simulation of the evolution of capi-tal ratios under stress (Figure 7.1).5 The chapter thereby seeks to provide answers to the following common questions in stress testing:

• How much do credit losses usually increase in the case of a moderate, medium, and severe macroeco-nomic downturn and/or financial stress event? For example, if cumulative real GDP growth turns out to be, for instance, 4 or 8 percentage points below potential (or average or previous years’) growth?

• How typically do other major factors that affect cap-ital ratios, such as profitability, credit growth, and risk- weighted assets (RWAs), react under these circumstances?

• Taking these considerations together, how does mod-erate, medium, or severe macro- financial stress trans-late into (a decrease in) bank capital, and thus, how much capital do banks need to cope with different levels of stress?

In answering these sorts of questions, a useful set of rules of thumb for stress testing should embody several properties:

• Coverage of the major factors contributing to banks’ vulnerabilities (in the first instance, in terms of solvency).

• Wide applicability, but with criteria to determine where inapplicable. A rule of thumb should be useful in many circumstances and many countries, but it should be clear where it should not be used.

• Robustness, implying that the rule is supported by a variety of evidence and not subject to excessive model risk.

• Intuitiveness, so that it can be used to interpret re-sults and inform decision- making.

As a corollary to these properties, a desirable rule of thumb should be relatively simple. Simplicity is likely to en-hance wide applicability, robustness, and intuitiveness. Rules will be developed along these principles.

To find common patterns, the study investigated various pieces of empirical evidence, including descriptive statistics, which may capture stress that does not necessarily originate from measured macroeconomic factors. The analysis in-cludes data from various previous crises, but focuses on the crises over recent decades (including the Russian/Asian cri-sis, crises experienced by the transition countries in central

on the data available, it may be wise to “import” a rule of thumb.

• When “model uncertainty” is a consideration: Even when national experience and data allow the con-struction of a relatively complex model that captures well past behavior, it could be less relevant in the fu-ture (for example, because certain asset classes are more or less relevant than they were in the past), and thus could give a false sense of accuracy; “model un-certainty” is an important consideration in stress testing and risk management more generally, though it is easy to overlook. Unless one knows with some precision the behavioral relationships that are rele-vant going forward, a simple rule of thumb may pro-vide the most reliable estimates and basis for action (Haldane 2012, 2013).3 Rules of thumb can be used also to check the plausibility of estimates of behavior derived from national experience.

• To test nonlinearities: It is often important to assess, and if possible quantify, the potential impact of non-linearities; stress tests are much more informative and credible if one can say how sensitive the results are to changes in assumptions (Taleb and others 2012). To this end, the ability to conduct multiple “runs” at low marginal cost using rules of thumb, rather than reanalyzing data in fine detail, is valuable to supervisors and managers.

• To assess various prudential rules: Rules of thumb can be used to assess the stability implications of various prudential rules, such as minimum capital require-ments, including on a cross- country basis. With rules of thumb, one can generate rough projections of the magnitude of shocks that banks typically face and what they could withstand, depending on their capitalization and other characteristics. Such projections would not replace detailed impact stud-ies, but would provide a plausibility and robustness check.4

• To assist decision- makers: Stress test frameworks (meth-ods and assumptions alike) and outcomes have to be readily accessible for senior managers and policymak-ers if action is to be triggered. A very complex model may be difficult to interpret and to link to policy instruments, and therefore it may distract from an in-formed debate on what actions should be taken; debate over the model may obscure debate over policy. To this end, decision- makers would be helped by the availability of readily understandable benchmarks

3 The possibility that “simple heuristics” may perform better than rigor-ous optimization in a world of uncertainty is now more widely acknowl-edged (see, for example, Gigerenzer, Hertwig and Pachur 2011).

4 A related paper on this is Dagher and others 2016, in which the authors study banking crises since 1970 and try to determine how much capital banks would need to avoid creditor losses and to minimize the portion of public recapitalization, but without looking at the drivers as done herein.

5 Effects of managerial action, such as the raising of capital, asset dispos-als, and balance sheet restructuring, and those of structural changes, such as exit of firms, are not analyzed here, in keeping with the standard methodology of stress testing.


Daniel C. Hardy and Christian Schmieder 121

severe loss levels are many times higher than in normal times, and in such circumstances banks typically exhibit substantially lower pre impairment income that can be used as a buffer against losses. In response to stresses, they re-strain from paying dividends or deleverage, the latter being a powerful way to restore bank solvency but a macroeconomi-cally costly alternative if it comes along with constrained credit supply.

It is found also that the interpretation of capital levels should take into account differences in the measurement of regulatory capital: the measured risk- based capitalization of a bank using the standardized approach (StA) under the Ba-sel II (and revised Basel III) standard will be less sensitive to both positive and negative shocks than that of a bank using the internal- ratings- based (IRB) approach.7 On this basis, one may suggest that the StA may be slow to reveal emerging vulnerabilities, while the IRB approach more quickly reflects deterioration in a bank’s situation (and rebounds faster when conditions improve), provided that changes in risk are re-flected in IRB risk weights on a timely basis.8 By the same token, a given level of risk- based capitalization during be-nign times may be less of a buffer for a bank using the IRB approach than for a bank using the StA.

The structure of the chapter and the main variables inves-tigated are as follows: Section 2 explains the main elements

and Eastern Europe in the late 1990s, the burst of the inter-net bubble in the early 2000s, the global financial crisis, and country- specific as well as bank- specific crises).

The evidence justifies distinguishing between emerging market economies (EMEs) and advanced economies (AEs) because their typical behavior differs importantly.6 These differences can plausibly be attributed to differences in mac-roeconomic performance— for example, EMEs tend to have relatively large cyclical fluctuations and more concentrated economies— and institutional factors, such as the effective-ness of loan workout mechanisms. Because the rules of thumb differ across AEs and EMEs, as do the typical magni-tudes of shocks, different levels of capitalization are needed to achieve a given level of resilience against potential shocks. Evidence from low- income developing countries (LIDCs) is used whenever possible, but less is available, and the func-tioning of LIDC economies may differ from that of both AEs and EMEs, for example, because of greater dependence on export of commodities and a much lower level of bank intermediation.

The evidence suggests also that the behavior of relevant variables (credit loss rates, income, credit growth, RWAs, capital ratios) is highly nonlinear around crises. Effects on bank capitalization and loan quality under a severe crisis are disproportionately great. In case of credit losses, for example,

6 Distinguishing between advanced, emerging market, and low- income economies is common practice in academic literature (such as Hardy and Pazarbasioğlu 1999), and policy- oriented analysis (such as the IMF’s Global Financial Stability Reports), in recognition of the impor-tant differences among them in terms of economic institutions, trends, and vulnerabilities.

How do average banks’ capital ratios change under stress?

Base scenarios onpast evidence.

Descriptive rules of thumb(Section 3):

Credit loss rate Probability of default Loss given defaultPreimpairment income Trading incomeCredit growth

Rules of thumb applied to typical banks (Section 5):

Total assetsRisk-weighted assetsProfitabilityCapital ratios Risk-based regulatory capital Leverage

Link bank solvency parametersto macro conditions.

Rules of thumb for satellite models(Section 4):

Credit loss rates Probability of default Loss given defaultPreimpairment income Retained earningsCredit growth Asset correlations

Source: Authors.

Figure 7.1 Formulating Rules of Thumb

7 This statement will remain adequate under the revised Basel III stan-dards for credit risk.

8 The risk sensitivity of RWAs under the IRB approach depends on the degree to which a bank’s rating system is based on “ point- in- time” rather than “ through- the- cycle” parameters.



growth in the balance sheet and specifically that of loans; and (2) risk, that is, changes in risk weights due to the changes of the risk profile of the banks’ assets, especially for those banks using an IRB approach.13

Other stability metrics are available and often useful— the unweighted leverage ratio, for example, is widely re-garded as a robust, complementary indicator. The poststress leverage ratio can be decomposed as follows:

tt

t t

t

Projected Leverage Ratio =Initial Capital +Projected Retained Net Profit

Initial Assets +Projected Change in Assets+1+1

+1

tt

t t

t

Projected Leverage Ratio =Initial Capital +Projected Retained Net Profit

Initial Assets +Projected Change in Assets+1+1

+1

(7.2)

where “assets” take account of on- and off- balance-sheet items, such as credit lines, commitments, and guarantees. The de-nominator of this ratio can be affected in a stress scenario if allowance is made for on- and off- balance-sheet growth, including through the write-off of losses. Factors affecting the numerator are the same as those for the risk- based capital ratio.

Return on capital (ROC) and return on assets are indica-tive of a bank’s ability to recover from a capital loss (see IMF 2011c and 2016). Indeed, profitability is the first line of de-fense of any bank against credit and other risk. A sufficiently profitable bank can earn enough to restore its capitalization even in the face of substantial stress, either by attracting new capital with the promise of dividends or by retaining earn-ings. A bank with low profitability will be less able to recover from even a brief negative shock. The proposed rules of thumb are useful for projecting these metrics as well.

The rules of thumb are derived from the literature on banking crises and statistical evidence on the evolution of bank loan quality and quantity obtained from two datasets:

• Long- sample evidence is provided by data on default rates for firms (mainly from the United States) over a period of 95 years (1920–2015) for AEs, and includes mainly Moody’s 201614,15 data. That series provided by Moody’s reports annual exposure- weighted his-torical default rates. This long sample includes five periods of substantial stress.

• Cross- country evidence is provided by data on the sample of banks available in Bankscope. The time dimension is limited to the period from 1996 to 2011, with the number of banks increasing in the later part of the sample. This period covers various episodes of stress in the countries covered. Evidence is obtained on bank performance in AEs, EMEs,

of the approach and the evidence available. Section 3 con-tains the proposed descriptive rules of thumb based on styl-ized facts of financial crises, where the rules are conditional on the evolution of credit losses. Section 4 contains rules- of- thumb versions of satellite models that link the key drivers of bank solvency— notably credit losses, preimpairment in-come, credit growth, and RWAs— to the evolution of GDP. Application of these rules to stylized banks based in an AE or an EME serves to illustrate their use and can be seen in Section 5. Section 6 concludes.

2. METHODOLOGY AND SOURCESThe main metric used in bank solvency stress tests is the capi-tal ratio, and especially the risked- based capital ratio defined as a bank’s capital divided by its RWA.9,10 The tests are meant to yield projections of capital ratios after stress, over the rele-vant time horizon. Hence, the rules of thumb described here are those most relevant to this objective. Post- shock risk- based capitalization can be decomposed as follows:

tt t

t t

Projected Capital Ratio =Initial Captial +Projected Retained Net Profit

Initial RWA +Projected Change in RWA+1+1

+1

d Capital Ratio =Initial Capital +Projected Retained Net Profit

Initial RWA +Projected Change in RWA+1+1

+1t

t t

t t

(7.1)

where the projected retained net profit is negative if net in-come is negative, and otherwise depends on the assumption made on dividend payouts.11 The main factors affecting net income that are relevant for solvency tests include:12

• Loan loss provisions (and write-offs, from an ex- post perspective), which in turn depend on probabilities of default (PDs) and loss given default (LGD) rates, or on some other rule for categorizing nonperform-ing loans (NPLs) and making provisions on them.

• Preimpairment income, including net interest in-come (including funding costs); commission and fee income; trading income; other operating income; and operating expenses.

• Dividend payouts and taxation. Both the numerator and the denominator of capital ratios

matter, especially if the projection period for a stress test is extended beyond a year. RWAs might evolve in ways that strongly affect the need for capital and a bank’s ability to meet regulatory requirements. The change in RWAs depends on two main factors: (1) volume, that is, the projected net

9 The capital ratio is called the “capital adequacy ratio” for regulatory purposes.

10 The Basel framework distinguishes between total capital, Tier 1 capital, and Core Tier 1 capital, the latter being made up by equity (common shares) and retained profit only.

11 For simplicity, it is commonly assumed that all effects of shocks go through the profit- and- loss account, rather than being taken out of capital directly.

12 And default rates, if one takes an ex- post perspective.

13 For the banks under the StA, changes in risk (due to changes in external ratings) will affect only the externally rated part of the credit portfolio, which is usually limited in size.

14 The majority of counterparts rated by Moody’s are based in the United States, especially in the earlier part of the sample.

15 This database is the longest readily available and relevant time series for bank solvency research.



3. TYPICAL BANKING CRISES AND DESCRIPTIVE RULES OF THUMBA simple but possibly robust approach to defining and cali-brating a stress test for a banking crisis is to look at historical episodes, not just in one country— which may have limited experience— but in a range of comparable countries. Past crisis episodes and severe recessions provide highly relevant information on the impact of stress on bank solvency through various channels. In what follows, the chapter seeks to come up with descriptive rules of thumb, after some sug-gestive conclusions on the link between macroeconomic conditions and banks’ performance.

Literature on Banking Crises

There is a large literature on banking crises, mostly looking at macroeconomic precursors and effects, and at the effec-tiveness of the strategies adopted on different occasions.17 A subset of studies provides quantitative evidence that is of direct relevance to stress testing. One example thereof is Čihák and Schaeck (2007), who look at 51 episodes of bank-ing crisis during 1994–2004, in a sample that covers coun-tries from every region. Figure 7.2 illustrates the typical behavior of default rates through a crisis, showing the evolution of the stock and (a proxy for) the inflow of NPLs relative to total loans three years on either side of a crisis peak. Table 7.1 provides more detailed estimates, broken down by region.

It can be seen that NPL stock ratios (in many countries still the most commonly available credit risk indicator) in-crease substantially during crises, and typically peak one year

and some LIDCs. The data covers more than 16,000 banks in 200 countries and jurisdictions, but the ma-jority of banks (more than 13,000) are based in AEs (almost half in the United States), about 3,000 are in EMEs, and only 550 are in LIDCs. There is some se-lection bias toward larger banks, especially in the LIDCs, which renders the evidence for these coun-tries less robust. For the establishment of the rules of thumb, outliers were removed from the sample to avoid misleading results, and other robustness checks were performed, leaving a sample of more than 10,000 banks from almost 170 countries.16 Summary statistics are provided in Appendix 7.1.

Such evidence drawn from a wide range of times and countries no doubt suffers from large differences in defini-tions of relevant variables, and the analysis of the evidence needs to recognize and cope with this challenge. Even today among countries using accounts based on International Fi-nancial Reporting Standards, criteria for classifying assets or defining capital, for example, can differ widely. Loss recog-nition and provisioning practices can differ from bank to bank within the same jurisdiction. Yet, for the purposes of this study, using a wide range of evidence is essential: only diverse data sources can yield evidence on the effects of ex-treme but plausible shocks, including shocks that force banks to reveal losses that are hidden in normal times. An attempt is made to limit the influence of data problems by using relatively robust techniques, and also by looking at in-dicators of both location and spread.

16 Data for some banks’ solvency variables contain numerous missing val-ues. All banks with less than five observations overall (during 1996–2011) were excluded from the sample. Very high loss levels (even above 100 percent) were observed for a few banks, for example, because of substantial off- balance sheet credit operation. Such outliers were re-moved from the sample.

17 Lo 2012, for example, provides a recent overview of studies related to macroeconomic and financial crises.

NPL stock ratio (mean)NPL stock ratio (mean + 1 StD)Proxy for default rate (mean)Proxy for default rate (mean + 1 StD)

0

35

30

3 –2 –1 0 1 2 3

Years before and after a crisis

NPL

stoc

k ra

tio a

nd d

efau

lt ra

te (p

erce

nt)

5

10

15

20

25

Source: Čihák and Schaeck (2007,15); and authors’ calculations.Note: NPL = nonperforming loan; StD = standard deviation.

Figure 7.2 Čihák and Schaeck Evidence on Typical Evolution of NPL Ratios Around a Crisis



EMEs tend to peak at higher levels, and are generally more variable than those in AEs. Also, real GDP growth is much more variable in EMEs. A useful set of rules of thumb needs to accommodate these major differences; different rules may be needed for AEs and EMEs. Similarly large differences in typical behavior across types of countries are documented in the analysis that follows.

Historical Evidence on Banking Crises

The long- sample evidence suggests broadly similar patterns: the default rates observed in the Moody’s data for firms dur-ing the last nine decades peak on five occasions, with the highest peak (unsurprisingly) occurring during the Great Depression of the 1930s (Figure 7.3).22

According to Giesecke and others (2011), default rates be-fore 1900 twice peaked at levels even higher than those seen during the 1930s (Figure 7.4). This finding may reflect the fact that the US economy before 1900 was more like that of an EME, being relatively highly dependent on commodity production and prices and prone to major infrastructure booms and busts (such as in railroad building). Also, arguably, economic policymaking and implementation were worse before the establishment of the US Board of Governors of the Federal Reserve System and major automatic stabiliz-ers. Hence, the historical United States corroborates the hy-pothesis that default rates are higher and more variable in EMEs than in AEs. The loss levels observed during the 1930s are kept as a reference point of an extreme AE crisis, with the caveat that conditions have changed substantially since then.

An implication of the evidence presented so far is that one should distinguish crises by their severity. There may be rela-tively mild crises that occur relatively frequently, and at greater intervals there may be major crises with more profound effects.23 A useful set of rules of thumb needs to accommodate these major differences; different rules may be needed for moderate, severe, and extreme stress situations. Specifically, this chapter distinguishes among:

• Normal conditions (in statistical terms, the median of credit loss rates)

• Moderate stress (the 80th percentile of credit loss rates)24

• Medium stress (the 90th percentile of credit loss rates)• Severe stress (the 97.5th percentile of credit loss rates)• Extreme stress scenarios (the historical maximum—

typically the worst year in a century).

22 The peak banking crisis year is defined here as the one with the highest default rate, which is consistent with the crisis definition used by Čihák and Schaeck (2007).

23 Similarly, Hardy and Pazarbaşioğlu (1999) investigate leading indica-tors of more or less severe banking crises.

24 The percentiles are estimated assuming that the annual credit loss rates are independent of one another. Yet, there is in fact some clustering of stress years, which implies that moderate and medium stress episodes, which can last several years, are less frequent than suggested by the percentiles. Episodes of moderate or medium stress normally occur in the AEs on a one- in- 10 to one- in- 15-year basis, and a one- in- 20-year basis, respectively.

18 The NPL stock ratio does not provide information on flows of credit losses, but instead provides proxies’ cumulative default rates less write- offs.

19 Insofar as a banking crisis is preceded by rapid credit growth, many credits do not mature until after the peak of the crisis and only then become “eligible” as nonperforming.

20 Fair value accounting and risk- based capitalization (under Basel II) are meant to remove the opaqueness of banks. Yet, regulatory rules are meant to follow a through- the- cycle approach, mitigating procyclical-ity. For stress testing purposes, point- in- time information is needed to monitor the current state of bank solvency. See the discussion in Section 5 of this chapter.

21 In case of less severe crises, the default rate is close to the long- term aver-age already two years after the trough of the crisis.

TABLE 7.1

Čihák and Schaeck Evidence on Typical Evolution of NPL Stock Ratios Around a CrisisRegion Number

of Countries

Change in NPL Stock Ratio (percentage

points)1

Output Loss (t + 1,

percentage points)

Average StD Average StDEmerging

countries2

12 12.7 9.6 −4.9 6.2

Asia 4 16.4 13.5 −7.8 5.4 Europe 4 8.7 6.2 −0.1 5.8 Latin

America4 11.8 8.9 −6.8 5.6

Advanced economies

5 3.1 0.7 −0.8 2.1

All countries (FSAP)

17 9.6 9.1 −3.7 5.6

Source: Čihák and Schaeck 2007.Note: FSAP = Financial Sector Assessment Program; NPL = nonperforming loan; StD = standard deviation.1Percent of loans overdue 90 days or more.2Country sample “Asia“ includes Indonesia, Korea, Thailand, and Philip-pines; “Emerging Europe“ includes Czech Republic, Russia, Slovak Repub-lic, and Turkey. “Latin America“ includes Argentina, Brazil, Mexico, and Uruguay.

after the materialization of a crisis.18 The temporal shift re-flects the fact that some loans default with some time lag after the materialization of macroeconomic stress, and that many banks have tended to recognize NPLs with delay, that is, they do not provision fully all potential losses after the first year(s) of a crisis.19,20 The NPL stock ratios rise by about 10 percent-age points from the typical level one year before the crisis in “average” crises, and almost 25 percentage points in severe crises. The stock is persistent: even after three years, the NPL ratio is at about the same level when the crisis materializes.

Using NPL flow ratios as a proxy for default rates (Box 7.1), the chapter finds that they peak at about 10 percent in “average” crises, and at about 18 percent in severe crises, up from 3 percent in “normal” times. The evidence here suggests that default rates come down to precrisis levels after three years, which makes their pattern roughly symmetric with re-spect to the crisis. The time needed to resolve problem loans implies that reduction in the stock tends to take longer.21

One immediate implication of this evidence by type of country is that the NPL ratio and the NPL flow ratio in



Box 7.1. Proxies for Credit Loss Rates

For analytic purposes it is often most useful to consider probability of default (PD) and rate of loss given default (LGD), the product of which equals the rate of credit losses per time period. However, information on PDs and LGDs is often unavailable, and so proxies have to be used. Given that some of these proxies capture stocks rather than flows, ancillary assumptions are needed to estimate those rates.

Banks’ profit and loss accounts provide evidence on (past) credit loss rates. Generally, accounts are meant to recognize potential losses when there is firm evidence of impending default, and value them depending on expected losses given default. Accounting practices dif-fer in the extent to which they are forward looking and the discretion banks have in recording the timing and magnitude of losses.

Many countries report data on nonperforming loan (NPL) stocks. The PD during a certain period from t−1 to t, in a forward- looking con-text, can be approximated by PDt = (NPLt − NPLt−1) + αNPLt− 1, where α is the portion of NPLs that is written off (or otherwise resolved) during that period. Here, for the years before the crisis, α is set to 0.5, which means that three quarters of NPLs are fully written off after two years. For the period after the peak of NPLs, it is assumed that loans are dealt with at a slower pace, with α = 0.33 implying that about half of NPLs are resolved after two years. In practice this pace is affected by differences across countries in the definition of NPLs and institutional provi-sions for resolution, and by shifts in the severity of impairment of NPLs, which would normally be reflected in provisioning rates.

0

20

1880 1900 20 40 60 80 2000

Valu

e–w

eigh

ted

defa

ult r

ate

(per

cent

)

5

10

15

Source: Giesecke and others 2011.Note: The figures are based on value- weighted default rates, while the data from Moody’s (2013) focuses on the percentage of issuers that default.

Figure 7.4 Historical Corporate Bond Default Rates (1866–2008)

Real GDP Growth (RHS)One-Year Default Rate (All grades, LHS)

0

9

–15

25

Defa

ult r

ate

Real

GDP

gro

wth

(yea

r ove

r yea

r)

1

2

3

4

5

6

7

8

1920 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 2000 05 10

–10

–5

0

5

10

15

20

Medium

Severe

Moderate

Sources: Default Rates: Moody’s 2016; Real GDP Growth: Federal Reserve Bank of St. Louis.Note: LHS = left- hand side; RHS = right- hand side.

Figure 7.3 Historical Annual Default Rate for All Rating Grades (Percent)



This evidence confirms the previous observation that credit loss levels typically have been substantially lower in AEs than in EMEs and LIDCs, while loss levels are found to be quite similar for EMEs and LIDCs. For AE banks, credit loss rates peak roughly at 0.8 percent under moderate stress, 1.1 percent for medium stress, and 2.4 percent under severe stress conditions, while levels of more than 3 percent are the historical maximum levels for country aggregates. Across stress levels, losses are higher in EMEs by a factor of about three compared to those in AEs. Yet, Figure 7.5 shows that this is a general pattern, and that the results for the three country groups overlap, that is, that there are countries with EMEs where banks have experienced lower median loss rates during the past 15 years than some of the countries with AEs. The same also holds true for moderate, medium, severe, and extreme stress levels— an important reason for this re-sult being that, during the sample period of 15 years, only some of the countries experienced a banking crisis. The fact that the means exceed the median even in normal times in-dicates that there is a “tail” of countries with much higher loss rates, a phenomenon apparent from the figure.

Comparable results are obtained from bank- by- bank data. The maximum credit loss level for each bank was as-signed to the respective level of severity.25 To assess the full pattern of solvency parameters around crises, seven consecu-tive observations are used: three before the peak of the crisis, the peak year, and three afterward. Given the rather limited length of the time series of the Bankscope dataset (Appendix Figure 7.1.1), just one severity level (the maximum) per bank

Source: Authors, based on Bankscope data.Note: AE = advanced economies; EME = emerging market economies; LIC = low-income countries.

Figure 7.5 Median Loss Rates by Country (1996–2011) (Percent of credit outstanding)

0.0

7.0

1.0

2.0

3.0

4.0

5.0

6.0

AE EME LIC

25 For AE banks, those with maximum credit loss levels between 0.4 per-cent and 1 percent of assets were deemed to have undergone a moderate stress scenario. For medium level losses, AE banks with maximum loss rates between 1 percent of 2.4 percent were deemed to have undergone medium- intensity strain. All AE banks with loss levels above 2.4 were deemed to have been subject to a severe/extreme scenario (there were not enough observations to distinguish severe from extreme episodes). Banks from EMEs and LIDCs were similarly categorized, albeit with different definitions of the severity of crises (Table 7.2).

TABLE 7.2

Typical Credit Loss Rates under Different Levels of Shocks(Medians except where indicated; percent of credit outstanding)

Scenario Credit Loss Rates

AE EME LIDCNormal, median 0.3 1.0 1.4Normal, mean 0.7 1.9 2.4Moderate 0.8 2.2 3.1Medium 1.1 3.4 4.9Severe 2.4 7.4 13.4Extreme 4.3 14.0 33.8Number of observations 463 1,457 698Number of countries 32 104 52

Source: Authors, based on Bankscope data.Note: AE = advanced economy; EME = emerging market economy; LIDC = low-income developing country.

Applied to the long- term evidence on default rates from Moody’s, one of the five crises would be classified as “moder-ate/medium,” two as “medium,” the global financial crisis as a borderline case between “medium” and “severe,” and the crisis in the 1930s as a “severe/extreme” crisis (Figure 7.3).

Descriptive Rules of Thumb

Credit loss rates

The recent cross- sectional evidence (from 1996–2011) on the effects of crises on banks’ loan quality corroborates the long- term evidence. Table 7.2 shows the annual credit loss rates (flow of loan loss provisions from profit-and-loss accounts relative to the total loan stock) banks should expect under various levels of stress, distinguishing between AEs, EMEs, and LIDCs. These parameters have been determined based on the median annual loss rates per year; the historical per-centiles are computed for each of the three country types corresponding to the respective stress level, with a view to allowing for a wide comparison.



was included, which filter also precludes the multiple use of data points.

Banks with very low loss levels or extremely high loss lev-els (loss levels above 60 percent, which applies to banks with unusual business models such as public banks involved in credit guarantee business) were excluded from the analysis to avoid distortion (Appendix 7.2). As a means to test robust-ness, estimates were recalculated using a sample including only those banks with at least five consecutive observations during the stress years (from –2 to 2 years); results were qualitatively similar.

Credit loss rates are found to peak sharply during one single year, and the credit loss rate pattern is symmetrical with respect to the crisis (Figure 7.6 and Appendix Table 7.2.1).26 This finding is consistent with other results, but it comes out more clearly based on the large sample of banks, and suggests that credit loss rates from the profit-and-loss accounts are indeed a useful proxy for banks’ default rates and LGDs.

“Normal” Moderate Medium Severe

0.0

4.5

Years around crisis

1. Advanced Economies

Years around crisis

2. Emerging Market Economies

0.0

18.0

Years around crisis

3. Low-Income Developing Countries

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

–3 –2 –1 0 31 2

–3 –2 –1 0 31 2

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

0.0

18.0

–3 –2 –1 0 31 2

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

Source: Authors, based on Bankscope data.

Figure 7.6 Typical Evolution of Credit Loss Rates Under Stress (Median loss rate by stress severity, percent of credit outstanding)

26 Use of medians rather than means should contribute to robustness against outliers.



Figure 7.8 shows that LGD and default rates are posi-tively correlated (the pairwise correlation is 0.47), but LGD rates tend to fluctuate proportionately less than do default rates and are less cyclical.29 The relationship allows one to project LGD under stress scenarios of various degrees of se-verity, where the severity of the scenario is captured by the respective default rate (Table 7.3). LGDs typically double in the case of a severe/extreme stress, whereas default rates in-crease many times over, albeit starting from very low levels in normal times. However, because LGD variations have a linear impact on RWAs, their impact on bank solvency is still considerable. Moreover, in absolute terms, LGDs greatly amplify stress under severe conditions.

Loss given default

There is empirical evidence that LGDs also fluctuate with the business cycle; note that in Figure 7.7, the circled stress years are the same as those in Figure 7.3. Descriptive evi-dence for workout LGDs (that is, LGDs for bank loans) from Moody’s 2016 is used to determine changes of LGDs under stress for AEs. The time series by Moody’s for loans, which relate to industry averages rather than bank- by- bank results, spans the period from 1990 to 2015.27,28

27 A longer series for bonds is available.28 Broadly similar results are obtained in the study by Araten, Jakobs, and

Peeyush 2004, which is uses a sample of more than 3,700 defaulted loans (predominantly US exposure) and spans the period of the savings and loan crisis, and in that of Caselli, Gatti, and Querci 2008, which uses a sample of over 11,600 Italian bank loans.

Real GDP Growth (US) (RHS) Araten, Jakobs, and Peeyush (2004) (LHS)Moody’s (2016), Loans (LHS) Moody’s (2016), Bonds (LHS)

0

90

–4

8

LGD

Real

GDP

gro

wth

(y-o

-y)

1982 12

–2

0

2

4

6

10

20

30

40

50

60

70

80

85 88 91 94 97 2000 03 06 09

Source: LGDs: Araten, Jakobs, and Peeyush 2004; Moody’s 2016; Real GDP Growth: Federal Reserve Bank 2016.Note: LGD = loss given default; LHS = left- hand side; RHS = right- hand side; y- o- y = year over year.

Figure 7.7 Evolution of LGDs through the Cycle (Percent of the face value of affected credits)

0

50

0 1 2 3 4 5 6 7

Default Rate

LGD

5

10

15

20

25

30

35

40

45

y = 3.5068x + 23.669R 2 = 0.2257

Source: Authors, based on Moody’s 2016 data.Note: LGDs = losses given default.

Figure 7.8 Link between LGDs and Default Rates (Percent)

29 Background information on Figure 7.7 can be found in Schmieder, Puhr, and Hasan 2011.



Preimpairment income

The next question is how preimpairment income evolves when stress occurs, that is, conditional on credit losses.32 The key components of preimpairment operational income are (1) net interest income, which tends to be relatively stable insofar as interest rate changes are passed on to loans and deposits when they mature; (2) net fee and commission in-come; (3) trading income; (4) other operating income; and (5) operating expenses. The remainder is other nonoperating income, which is normally neglected because it represents one- off items that do not affect the sustainability of a bank’s business model. In order to normalize the measure of in-come despite large cross- country differences in banks’ lever-age, the study focuses on the ROC rather than return on assets.

Using bank- by- bank data in the Bankscope database, median net preimpairment ROC is found to be highest in LIDCs (25.0 percent), followed by EMEs (18.9 percent) and then AEs (11.9 percent), in line with expectations and ac-counting for the rank order in terms of the risk of doing business as captured by the volatility of loan losses.

Preimpairment income conditional on credit loss rates is found to be affected in stress scenarios, but on average to remain positive and thus a buffer against credit losses (Figure 7.9 and Appendix Table 7.2.1). The behavior of in-come with respect to the crisis trough tends to be less sym-metric than that of credit loss rates; income often remains low for some years after the trough (for AE banks) or drops significantly at the time of the trough (for EME banks).33

However, there is substantial variation to this finding across banks. If one looks at the AE banks that are more ad-versely affected, that is, the worst performing 25th and 10th percentiles, income becomes negative for at least a quar-ter of banks under severe stress (Figure 7.10); 10 percent of banks encounter ROCs below –20 percent at the time of the trough under severe stress. A similar pattern can be detected for EME banks, but the limited quantity of available data precludes firm calibration.

Certain earning sources seem especially vulnerable to stress conditions, as illustrated in Figure 7.11 (with detailed data in Table 7.5), which shows the standard deviations across crisis periods of the median ratio to capital of various income and expense account components. For AE banks, net commission and fee income is consistently relatively volatile. The median advanced economy bank’s operating expenses become very volatile in a severe crisis, perhaps because a severe crisis will force a bank to bear restructuring costs, and because capital is reduced. For the median EME bank, the change of net in-come under stress is largely driven by net interest income due to foregone interest on credit losses and, possibly, interest rate

32 Net preimpairment income is generally referred to, that is, adjusted for operational expenses but not loan losses, unless stated otherwise.

33 The outcomes for LIDCs are shown in Appendix Table 7.2.1, but evi-dence is very limited.

30 The chapter assumes that the LGDs reported by the World Bank survey (covering 181 countries; see http://www.doingbusiness.org) are a proxy for LGDs for corporate exposure. To account for lower LGDs on mort-gages, a retail LGDs of 25 percent for Organisation for Economic Co- operation and Development countries is assumed, 45 percent for countries with emerging markets, and 50 percent for LIDCs. It is also assumed that 40 percent of total credit is retail and corporate LGD rates for the remaining credit are applied. For AEs, a floor of 30 percent is assumed for corporate credit, accounting for findings in Schmieder and Schmieder 2011. The latter study investigated recovery rates conditional on legislation, and found drivers relating, for example, to legal proce-dures that account for the wide range of recovery rates.

31 If one divides the credit loss rates observed for advanced countries in Table 7.2 by the LGD for the United States (0.42), the implied default becomes 0.7 percent (normal times, the global median), and 1.9 per-cent, 2.6 percent, and 5.7 percent for moderate, medium, and severe stress, respectively. This compares to equivalent default rates (Table 7.4) based on long- term data from Moody’s of 0.7 percent (normal), 1.9 per-cent (moderate), 2.9 percent (medium), and 5 percent (severe).

TABLE 7.3

Stress Levels of Default Rates and LGDs for AEs(Percent)

Scenario Normal Moderate Medium Severe ExtremeDefault rates1 0.7 1.7 2.9 5.0 4.4Projected LGD 26.0 30.0 34.0 41.0 54.0

Source: Authors based on Moody’s 2016 data.Note: AEs = advanced economies; LGD = loss given default.1Based on Moody’s default rates shown in Figure 7.3.

LGDs for EMEs and LIDCs are not covered due to lack of data. The historical average LGD rates for EMEs and LIDCs are about 59 percent and 62 percent, respectively. These LGDs have been derived from the World Bank (Doing Business) and are assumed to correspond to long- term aver-age LGDs.30 A practical approach for stress testing purposes would be to use the long- term average level in a specific country and, for a stress scenarios of given intensity, add the same absolute increase as seen in AEs.

Credit losses (should) account for fluctuations in both PDs and LGDs (Box 7.1). One may therefore wish to compare the default rates in Table 7.3 with the loss rates in Table 7.2 for the AEs by dividing the loss rate by the respective LGD for the stress level. The median long- term default rate observed in Moody’s data (0.7 percent) translates into a loss rate of about 0.29 percent (using the LGD for the United States, as deter-mined by Schmieder and Schmieder 2011, of 0.42). This esti-mated loss rate is comparable to the 0.24 percent median loss rate for the US banks in the Bankscope data (for the period covered, that is, 1996–2011).31 More generally, if one divides the credit loss rates in Table 7.3 by the LGDs in Table 7.4, the implied default rates are 1.3 percent (normal), 2.7 percent (moderate stress), 3.3 percent (medium stress), 5.7 percent (severe stress), and 8 percent (extreme stress). On the lower end, the implied default rates are higher than the actual default rates, unless one replaces the projected LGDs by some average LGD for advanced countries (35 percent). Overall, this result suggests that credit loss rates from the profit-and-loss accounts, divided by LGD rates, provide reasonable approximations to PDs, at least for AEs. The same applies to the implied default rates for EMEs and LIDCs using an LGD of 60 percent.


http://www.doingbusiness.org


behavior during crises.34 EME banks manage to reduce oper-ating expenses at times of stress, unlike AE banks.

A common perception that trading income is only mildly correlated with the cycle is generally true except in severe crises, when trading income can become highly negative (Figure 7.12).35 The likelihood of sizable trading losses is dis-cussed further in Box 7.2, highlighting that “black swan” events need to be captured by stress tests for banks with meaningful trading operations.36

Dividend payout is found to drop to zero under severe stress, for AE, EME, and LIDC banks (Appendix Table 7.2.1). Tax payments are also reduced, reflecting lower net income.

Credit growth

Banking crises affect not only the quality of a given stock of loans, but also asset and loan growth rates; nominal credit growth of banks’ customer loans net of credit losses tends to slow sharply, even if it does not become negative (Figure 7.13 and Appendix Table 7.2.1).37 Conditional on stress (as measured by credit losses), deleveraging occurs only in case of severe crises in AEs, while medium intensity crises tend to end in three years of (close to) zero year- over- year credit growth for AEs. Credit growth remains fairly sizable for EME banks under stress, except for severe stress, when credit growth becomes zero. Again, the pattern for the LIDC banks is less clear- cut, owing to limited data. As-set growth behaves very similarly to credit growth. Thus, both the risk- weighted capital ratios and the unweighted leverage ratio may be affected by this effect in the denominator.



Years around crisis

10.0

28.0

Years around crisis


13.0

16.0

19.0

22.0

25.0

7.0

15.0

–3 –2 –1 0 21

–3 –2 –1 0 21

9.0

11.0

13.0


Figure 7.9 Typical Evolution of Preimpairment Income Under Stress (Median preimpairment ROC by stress severity; percent)

34 Plausibly, many EME banking crises are associated with balance- of- payment crises, which may lead to higher short- term interest rates, whereas central banks in AEs react to financial sector pressure by reducing short- term rates and can afford to ignore balance- of- payment effects.

35 Trading income is included under “other operating income” in Fig-ure 7.15, which shows results for medians. Trading income is relatively unimportant and stable for the median bank, which is quite small.

36 See Taleb 2010, and Taleb and others 2012.

37 This study, including in Figure 7.13, investigates nominal credit growth rates, and is therefore consistent with the other factors affecting sol-vency, which are likewise measured in nominal terms.


Source: Authors, based on Bankscope data.Note: ROC = return on capital.

Figure 7.10 Evolution of Preimpairment Income for Worst Performing Banks under Stress (Preimpairment ROC by stress severity, percent)


1. Advanced Economies (Lowest 25th percentile)

Years around crisis

0

25

Years around crisis

3. Emerging Market Economies (Lowest 25th percentile)

5

10

15

20

–4

14

–3

–3

–2 –1 0 2 31

–2 –1 0 2 31

0

4

8

–2

2

6

10

12


Years around crisis

–10

25

Years around crisis

4. Emerging Market Economies (Lowest 10th percentile)

0

–5

5

10

15

20

–25

15

–3

–3

–2 –1 0 2 31

–2 –1 0 2 31

–20

–10

0

–15

–5

5

10

Moderate Medium Severe

0.0

5.0

Net interestincome

Net preimpairmentincome

Fees Other operatingincome

Operatingexpenses

0.0

5.0

Net interestincome

Net preimpairmentincome

Fees Other operatingincome

Operatingexpenses

1.0

2.0

3.0

4.0

1.0

2.0

3.0

4.0




Figure 7.11 Standard Deviation across Crisis Periods of Median Income and Expense Components (Percent of capital)



Box 7.2. How Likely Is It That Large Trading Losses Coincide with Large Credit Losses?

In many stress testing exercises in the past, no explicit link could be established between credit losses and trading income. And there is a good reason for that, as shown in Figure 7.12 for a median bank with some trading operations. Trading income is, on average, slightly posi-tive under most stress conditions. However, this is not necessarily the case if one moves further into the “tail,” and in particular if one fo-cuses on the experience of banks with sizable trading books under severe conditions. Looking at the severe scenarios as defined by credit risk losses, a bank at the worst performing decile (in terms of the ratio of trading income losses to capital) could lose about 5 percent of capital. The worst performing 1 percent of banks might lose a third of capital— which would likely be more than a bank could suffer and still survive combined with other losses and reduction of income banks would likely face under such conditions. The severe scenario may occur with low probability, but the impact could be large, especially because trading activity is concentrated in larger banks: the worst performing 1 percent of banks could hold a substantial share of aggregate assets. The nonlinearity and concentration of trading losses was apparent during the recent global crisis, when a few large European banks lost within one year more than one third of their capital due to trading losses alone, while many smaller banks had insignificant trading losses.

Trading income is likely to follow close to a random walk pattern, with little serial correlation, so a series of very bad years are unlikely to occur, but the simulations outlined earlier in this chapter suggest that one bad year of trading could be enough to bring down a large bank.



Years around crisis

–12

0

Years around crisis


–10

–8

–6

–4

–2

–0.9

0.0

–3

–3

–2 –1 0 2 31

–2 –1 0 2 31

–0.7

–0.5

–0.3

–0.8

–0.6

–0.4

–0.2

–0.1


Years around crisis

–40

0

Years around crisis

4. Advanced Economies (Lowest 1st percentile)

–30

–35

–25

–20

–15

–5

–10

–6

0

–3

–3

–2 –1 0 2 31

–2 –1 0 2 31

–4

–5

–3

–2

–1


Figure 7.12 Trading Income under Stress, by Quantile (Percent of capital)



It should be noted that satellite models are best suited to capturing the effects of exogenous macroeconomic shocks, rather than shocks that originate within the financial sys-tem, for example due to asset bubbles or large- scale malfea-sance (Alfaro and Drehmann 2009).38 In the latter situation, the macroeconomic deterioration tends to be a consequence of, and to follow the financial sector disturbance. Hence, the time series properties of the satellite models should be sensi-tive to the origins of the disturbance.

Explanatory Variables and Estimation Approach

The most important single determinant of bank solvency is the overall conjunctural conditions prevailing in the econ-omy, which will affect all solvency factors (credit losses, in-come, credit growth, and so forth). Strong economic activity should generally allow firms to generate the revenue to repay loans, households to earn steady income to meet debt service obligations, and collateral to retain its value. Weak economic

4. RULES OF THUMB FOR SATELLITE MODELSAn alternative to basing stress tests on generic crisis scenarios (referred to as descriptive rules of thumb) is to project the impact of some major shocks to macroeconomic variables on banks’ solvency variables. This more complex but more in-terpretable approach relies on so- called satellite models that describe macro- financial linkages. Various types of satellite models have been used in the literature to establish such relationships, including time series analysis, linear and non-linear regression models (such as OLS regression, logistic regression, panel analysis), and structural models (see Drehmann 2009 and Foglia 2009, for examples).

For the purpose of this chapter, the attention focuses on fairly simple relationships, albeit with allowance for nonlinear responses that depend on the severity and duration of strain as measured by a shock to real GDP growth. More complex rela-tionships might be estimated where a stress tester has available a rich dataset, covering periods (but without structural breaks) with a variety of shocks for a variety of banks. Yet even in such circumstances, transparent, understandable rules of thumb may be useful in checking the robustness of results.



Years around crisis

0.0

35.0

Years around crisis


5.0

10.0

15.0

20.0

25.0

30.0

–6.0

12.0

–3 –2 –1 0 31 2

–3 –2 –1 0 31 2

–3.0

0.0

3.0

6.0

9.0

Source: Authors, based on Bankscope data.Note: AEs = advanced economies; EMEs = emerging market economies.

Figure 7.13 Typical Evolution of Credit Growth for AEs and EMEs Under Stress (Percentage change, net of credit losses)

38 Malfeasance seems to have been a major contributing factor in recent banking crises in the Dominican Republic and Afghanistan, for example.



These GDP growth shocks (for the respective country type— AE or EME) are compared to the behavior of the main variables related to bank solvency. The comparison is carried out by various means: simple ratios and correlations are calculated, as are regressions. The main estimates are based on bank- level and country- level evidence. Given that the sample of bank- level data is dominated by observations from a few countries (notably the United States), which re-sults in limited variation for the GDP trajectories, country- level data was used to come up with the rules.41 However, because then the evidence is limited to some 20 observations per country type and stress level, some parameters had to be smoothed, with a view to ensure consistency across esti-mates: the level of the macroeconomic shock (in terms of GDP growth) multiplied by the sensitivity of the solvency parameters (computed based on bank- and country- level data) added to the preshock level should result in broadly the same stress levels as observed under the descriptive rules dis-cussed above.42 While the unadjusted results achieved such consistency in qualitative terms, the median rules were ad-justed slightly to align with the respective descriptive rules of thumb.

It must be recognized that a given macroeconomic shock will have diverse effects across banks (and countries): some will survive quite well, and others may be devastated. Some banks may suffer large losses in calm periods. From a stabil-ity perspective, policymakers are likely to be as concerned about the “tail” of weak banks as they are about the mean or median bank; a systemic and macroeconomic problem can be created by severe losses in just one quarter or even one tenth of the banking system. Hence, the emphasis here is not on the median bank, but on the distribution of results across banks and in particular the weaker banks. It turned out that the tail sensitivities, taken from the computed country- level GDP sensitivities, deliver solvency parameters that align well with the descriptive data for the corresponding confi-dence level, and represent worst- case crisis elasticities.43

activity will have corresponding negative effects on a bank’s clients and thus on the bank itself.

Real GDP is normally the most relevant and most readily available measure of aggregate activity. Policymakers and others make frequent real GDP forecasts, and its relation-ship to other macroeconomic variables is well- studied. Hence, its behavior can be readily interpreted, and it can usually be forecast in both baseline and stress scenarios.39 The design of scenarios is one of the key challenges for stress tests, and running a number of potential scenarios allows studying sensitivities (Taleb and others 2012)—a key advan-tage of using rules of thumb is that they facilitate working through numerous scenarios. In what follows, rules of thumb will be established relating variables related to bank solvency to GDP growth.

Economic intuition and evidence presented earlier in this chapter suggest that, if economic conditions are broadly stable and as anticipated, then only a (very) small proportion of loans will go bad (Figure 7.3, Table 7.2). However, nonper-formance may increase rapidly if conditions are unexpectedly adverse, which often happens only after a long period of be-nign times, and makes such shocks all the more challenging; it is these negative “surprises” that cause borrowers to be unable to repay and collateral to be reduced in value. Yet, as displayed previously, what counts as an exceptionally large shock in, say, the United States or a western European coun-try, may be well within recent historical experience for many EMEs. Moreover, economic intuition and some empirical evidence suggest that prolonged periods of low or negative growth will have a more pronounced effect on bank profit-ability than a brief recession followed by recession.

With this in mind, the study computed the average (1) changes in real GDP growth at time t = 0 (the year with the lowest real GDP growth) relative to year t−4; and (2) the cumulative deviation of real GDP growth rates from trend from t−4 to t = 0, using World Economic Outlook (WEO) data for both AEs and EMEs.40 The cumulative deviation from trend is likely to be more telling (given that it contains information on the duration of the shock), but depends on an estimate of trend growth, which in some cases (for example, following an unsustainable burst of growth) may be difficult to obtain. A practical approach to determine trend growth is to use the average GDP growth observed in the past over at least one complete cycle (say, the last 10 years) or baseline forecasts (for example, data from the WEO). For this study, the average real GDP growth rate for 1980 to 2011 has been used as a benchmark.

39 In some countries, statistics on real GDP are subject to long lags or measurement error. An index of industrial production is often an alter-native available with shorter lags and at higher frequency, but real GDP is still normally the preferable explanatory variable because it captures a wider measure of activity.

40 Note that the peak of the macroeconomic crisis (t = 0) is defined by the low point of GDP growth. It will be investigated whether the macroeco-nomic crisis peak coincides with the banking crisis peak, as defined pre-viously as the year with the highest rate of credit losses.

41 The computation included only observations with a drop of GDP growth or negative cumulative deviation, respectively, that is, crises observations.

42 The implied sensitivity, that is, the sensitivity based on inferring the sensitivities from the “average” size of the macroeconomic shock and the level of the solvency parameters computed based on the descriptive rules, is slightly higher than the one computed as the ratio of change of solvency parameter and change of GDP growth rates. This discrepancy arises because macroeconomic stress and financial stress are not always temporally aligned, and because idiosyncratic factors at the bank level are also at play. The proposed rules using the implied sensitivities makes the rules more conservative, in line with the purpose for stress testing, and consistent with the descriptive rules.

43 The 5 th- tile of the elasticities in Table 7.4 represents a tail level sensitiv-ity, but does not correspond to the single most extreme cases observed in the past. Credit losses in Iceland, an advanced economy, peaked at well above the 4.3 loss rate, the “extreme” advanced economy level in Table 7.3. Using the first percentile for the credit loss sensitivity for the “drop in GDP” rule in Table 7.4, for example, would yield a sensitivity of −2.5, corresponding to an increase on the credit loss rate to more than 30 percent, as observed for Iceland.



stress) at the trough. The median advanced economy bank’s credit losses would increase by (−2.4) × (−0.2) = 0.48 percent-age points, (−4.3) × (−0.2) = 0.86 percentage points, and (−7.4) × (−0.4) = 2.96 percentage points, respectively.

Projections along the path from precrisis to peak crisis can be calculated. For the rule based on changes of annual growth rates from precrisis levels and assuming moderate stress, for example, one would establish the level of solvency parameters as follows: for t = –3, the assumption of an initial drop of GDP by 0.3 percent, say, would yield loss rates of 0.36 (0.3 percent plus (−0.3) × (−0.2). For the next year (t = −2), the same sensi-tivity (0.2) would be multiplied with the change in the growth rate and added to the previous year’s level, and so on.

For stress testing when the system is already under strain, one would first have to decide, based on expert judgment, which stress level might occur (using the benchmark figures in Table 7.2), and which year ( t− 3 to t+3) reflects the current situation. For example, if one assumes that a severe scenario will occur, and that one is already at t = −2, one would use the sensitivities of the severe scenario to simulate the trajec-tories going forward. As in the previous case, one would ob-tain an estimate by adding the change of growth rates multiplied by the sensitivity of the respective stress level to the actual level of the solvency variable.

The table reveals that in many cases the sensitivity in-creases with stress. For example, the sensitivity of credit loss rates to GDP declines is roughly the same in moderate and medium stress situations, but at least twice as high under severe stress.45 Hence, the rules are tied to the categorized GDP trajectories. Thus, it is misleading to take a parameter from a moderate GDP shock and use it to project the effects of a severe shock, for example. If one seeks to simulate a drop in GDP growth by 3 percentage points, say, one would likely use the parameters for the moderate scenario (as 3 percent is close to 2.4 percent), but could also pick the parameters for the medium shock to compute an upper bound. However, not all relationships exhibit pronounced nonlinearity, as wit-nessed by the linear relationship between the size of the GDP shock and the change in credit growth rates.

The sensitivity of GDP shocks is much higher for a “tail” of weak banks than for the median bank. Across variables of interest, the coefficient for the worst performing 10th percen-tile of banks is normally at least double, and sometimes many times that of the median bank. The difference for ROC is especially large.

Credit quality and credit loss rates

For credit losses, a consistent observation is that large output falls can cause at least a one- for- one increase in default rates, which relationship is evidenced both by the findings reported earlier (Figures 7.2 and 7.3 and Table 7.1) and by further analysis of data from Moody’s 2016 (Figure 7.14 and

Possible time lags need to be considered in making these comparisons. For both credit loss rates and credit growth, their trajectories are found to be fairly symmetric with re-spect to the crisis, and the highest credit losses and lowest credit growth levels tend to concur with the year of the low-est real GDP growth. Hence, the results reported refer to coincident effects. Because preimpairment income is less symmetrical with respect to financial stress (conditional on credit losses; Figures 7.10 and 7.12), changes of real GDP growth rates are compared with the subsequent changes of pre impairment income, with a view to err on the conserva-tive side. For AEs, the change in income is computed as the minimum income level observed between t = 0 and t = 3, minus the initial income level at t = −4. For EMEs, income tends to remain high until stress materializes; hence, the av-erage income in t = −3 to t = 0 is compared with the mini-mum income level after t = 0 (in Appendix Table 7.2.1).

While the data available for LIDCs were sufficient to come up with some meaningful descriptions of the typical behavior of aggregates in financial crises, comparatively little data were available to investigate macro- financial linkages. Hence, macro- financial satellite rules of thumb are not re-ported for LIDCs.

Rules of Thumb for Satellite Models

Table 7.4 summarizes the estimated real GDP growth sensi-tivities of credit losses, preimpairment income, and credit growth, distinguishing, as before, among moderate, me-dium, and severe stress (as measured by the real GDP growth paths), and between AEs and EMEs. The column “typical base level at t- 4” reports the typical starting point before a stress event occurs. The other columns report the parameters of changes with respect to that baseline.

The table provides two different sets of rules, based either on cumulative changes in growth rates or on the change in the annual growth rates from precrisis to crisis trough.44 Each approach might be useful for stress tests in certain circum-stances. For example, one first establishes a scenario in terms of cumulative deviation of real GDP growth rates from trend during the years up to the trough of the stress period (such as 5.9 percentage points for the moderate stress in an advanced economy), which deviation is multiplied by the correspond-ing sensitivity parameter— in this case leading to an increase in median bank credit losses of (−5.9) × (−0.1) = 0.59 per-centage points in the worst year. Projections based on changes in annual growth rates work similarly: for an advanced econ-omy, for example, one could simulate the impact of a drop of real GDP growth rate from before the recession (perhaps 2.4 percent, the average real GDP growth rate in AEs during 1980–2011), to 0.0 percent (the moderate scenario for AEs), to −1.9 percent (medium stress), or to −5.0 percent (severe

44 The cumulative deviations of GDP growth are about twice the level of the change in annual real GDP growth rates for the AEs, and about three times in the case of the EMEs, suggesting that macroeconomic crises in the EMEs are longer and deeper than in AEs.

45 Marcucci and Quagliariello 2009 provide more formal evidence in support of this point.



divided by a decrease in GDP growth levels of 6.4 percent-age points); without the observations during the 1930s (where the financial cycle was lagging behind the business cycle), the absolute sensitivity is about 0.6 (= 3.1/4.8). The evidence collected for the rules of thumb established in Table 7.4 suggests that there is typically (in 60 percent of the cases) no lag between the GDP trajectory and the credit loss trajectory; in about 20 of the cases the GDP trajectory leads by one year, and in the remaining 20 percent of the cases the GDP trajectory lags by one year.

As shown in Figures 7.3 and 7.14, and in Tables 7.1 and 7.5, the bulk of credit losses typically occurs at the time of

Table 7.5). In the period covered by the Moody’s data, all five peaks of default rates during the last 90 years were accompa-nied by a sharp economic downturn, suggesting that finan-cial cycles and business cycles are usually closely linked in terms of their timing.46 In simple terms, based on Table 7.5, the absolute sensitivity of default rates to GDP growth is about 0.5 (increase of default rates by 3.5 percentage points

46 Note that the US recession at the end of World War II was not associ-ated with a rise in default rates, for example, and in case of the Great Depression during the 1930s, the most substantial losses occurred after GDP growth had hit bottom.

TABLE 7.4

Rules of Thumb for the GDP Sensitivity of Key Bank Solvency Variables

Typical base level at t−4

Stress level


Rules based on cumulative deviation of real GDP from trend, from t − 4 to t = 0

Cumulative GDP growth rate (4 years) Change of GDP growth rate

AE 10.0 −5.9 −8.5 −13.9EME 17.9 −11.5 −19.5 −32.7

Credit loss rate GDP sensitivity of credit loss rate

MedianLowest 10th

percentile MedianLowest 10th


percentileAE 0.3 −0.1 −0.2 −0.1 −0.2 −0.2 −0.4EME 1.0 −0.1 −0.3 −0.1 −0.3 −0.3 −0.6

Preimpairment ROC GDP sensitivity of preimpairment ROC

MedianLowest 10th



percentileAE 11.9 0.1 1.5 0.2 1.0 0.4 1.3EME 18.9 0.0 1.5 0.1 1.0 0.2 0.8

Credit growth rate GDP sensitivity of credit growth rate

MedianLowest 10th




Rules based on change in real GDP growth rate from t − 4 to t = 0GDP growth rate Change of GDP growth rate

AE 2.4 −2.4 −4.3 −7.4EME 4.2 −3.4 −6.6 −13.0

Credit loss rate GDP sensitivity of credit loss rate

MedianLowest 10th



percentileAE 0.3 −0.2 −0.4 −0.2 −0.4 −0.4 −0.8EME 1.0 −0.4 −0.6 −0.4 −0.8 −0.7 −1.5

Preimpairment ROC GDP sensitivity of preimpairment ROC

MedianLowest 10th




Credit growth rate GDP sensitivity of credit growth rate

MedianLowest 10th




Source: Author’s calculations based on studies and sources mentioned in the text.Note: AE = advanced economy; EME = emerging market economy; ROC = return on capital.



LGD

The correlation between LGDs for loans from Moody’s 2016 with GDP real growth rates is 0.44, indicating that LGDs are lower during periods of higher output (see also Figure 7.7). The evidence on hand, however, suggests that the sensitivity of LGD rates to a real GDP slowdown is usually moderate in AEs, given that LGD even in normal times is typically in the 30 to 60 percent range; the sensitivity (in relative terms) is lower than that of PDs (Table 7.6).48 For EMEs, there is not enough evidence to come up with separate parameters, but the parameters for AEs could be used as a starting point.

However, these rules of thumb for LGDs (which are espe-cially low for moderate and medium stress scenarios), are un-likely to apply when the source of the shock is related to the bursting of an asset price bubble: the end of an asset price

the trough of the associated macroeconomic crisis; the time of the lowest GDP growth coincides with the year when loss rates peak sharply— an important finding when it comes to specifying rules- of- thumb satellite models, and in line with most publicly available satellite models. Default rates come back to precrisis levels (that is, the long- term average of 1.1 percent) after three years. However, in the case of the 1930s Great Depression, losses remained elevated for several years after the trough in GDP growth, and did not come down to the average until after six years.

This is in line with GDP elasticities found based on the sample data from Bankscope (Table 7.4), where one can clearly observe the nonlinear (convex) pattern of credit loss rates, with the sensitivities increasing sharply as one moves from moderate to severe stress. The table also reveals that the GDP sensitivity of loss rates in EMEs is considerably higher than that found in AEs, and becomes close to one- for- one or more under severe conditions.47

For default rates, Moody’s 2016 data are used to compute GDP sensitivities for AEs (Table 7.6), which turn out to be qualitatively similar to loss rates computed based on bank- by- bank data discussed earlier in this chapter. Again, the pa-rameters in the table are meant to be used together with the GDP trajectories for the AEs from Table 7.4. For projections based on the cumulative deviation of GDP growth from trend, the parameters should be halved.

47 The coefficients for moderate stress are in line with the research of Ce-rutti and others 2010, for example, who used a linear panel model for emerging Europe. If one uses nonlinear specifications and/or does not control for effects other than GDP growth (that is, measures the impact of a macroeconomic shock by means of changes in GDP growth alone), the coefficients tend to be higher, which indicates that the established coefficients are robust.

Mean real US GDP growth rate excluding 1930sMean default rate Mean default rate + 1 SDLong-term average default rate Long-term median default rate

–1

8

–3 –2 –1 0 1 2 3

Years before/after a crisis

0

1

2

3

4

5

6

7

Perc

ent

Source: Authors, based on Moody’s 2016 data. Note: SD = standard deviation.

Figure 7.14 Historical Evidence on Typical Evolution of Default Rates Around a Crisis (Percent)

48 LGD levels depend on such features as the efficiency of legal proceed-ings for loan workout, but differ also across sectors: typically LGD rates are high for unsecured consumer loans and loans to small- and medium- sized enterprises (BCBS 2005).

TABLE 7.5

Historical Evidence on Typical Evolution of Default Rates around a Crisis

Number of

Episodes

Change in Default

Rate1

Change in Real GPD Growth

Rate

(Percentage points, t−4 to t)

Average StD Average StDAEs (with 1930s) 5 3.5 1.6 −6.4 3.9AEs (without 1930s) 4 3.1 1.6 −4.8 1.6

Source: Authors’ calculations based on Moody’s and Federal Reserve Bank data.Note: AEs = advanced economies; StD = standard deviation.1The trough of the crisis is set by the year with the lowest GDP growth.



Credit growth

For credit growth, the rule foresees similar GDP sensitivities for AE banks across shock levels. The sensitivity in EMEs tends to decrease with stress intensity, which result may be influenced by the comparatively high credit growth rates, especially in EMEs during the last 15 years. Going forward, the trend might change as EME financial systems mature, which is why these rules should be used carefully.

In applying these rules of thumb, it will again be impor-tant to make allowance for the origins of the simulated financial crisis: a crisis provoked by an external macroeco-nomic shock is likely to be associated with relatively stable credit aggregates or only a more moderate slowdown. A crisis provoked by the bursting of a credit- financed asset price bubble or a lending boom is likely to be followed by a more contractionary path for credit aggregates.

Asset correlations

Asset correlations are one of the solvency parameters that have been studied less intensely, despite their important impact on capital ratios. Besides the scarcity of long- term data, one reason for the relative neglect may be that IRB asset correlations are determined conditional on the level of PDs, based on a cross- sectional rule that foresees that lower PDs are associated with higher asset correlations.51 Using either IRB asset correlations (or fixed asset correlations) may be appropriate for supervisory purposes because the approach avoids possible double counting of potential stress when losses have already materialized (and affect the numerator of capital ratios), while the denominator (RWAs) (still) foresees the potential for elevated potential losses. However, if one takes a multiperiod, stress testing view, this relationship could be misleading, as illustrated below.

To determine the fluctuations of asset correlations under macroeconomic stress, we refer to a study by Düllmann, Scheicher, and Schmieder (2008) based on a sample of data on large, predominantly western European corporations.52 During a period of six years (1997–2003), asset correlations fluctuated strongly, ranging from 4 percent to 16 percent, with a mean at 10.5 percent (Appendix Figure 7.2.1).53 We find a high GDP growth sensitivity of −2.7 (Appendix Figure 7.2.2), which im-plies that a fall of GDP growth by 1 percentage point leads to an increase of asset correlations by 2.7 percentage points.

These parameter estimates are meant to offer a simple rule, but further evidence, including for the latest years of the financial crisis, would be helpful. In particular, more evi-dence on the behavior of asset correlations in more extreme

49 Basel III has defined minimum regulatory rates for retention of income (“capital conservation”) (BCBS 2011, paragraph 131).

50 It is easy to show that, if a bank’s total assets tend to grow at a rate g and its return on equity tends to be r, then its leverage ratio will remain constant if it retains a proportion (g/r) of its net profits.

51 This rule reflects the empirical fact that the credit quality of larger firms exhibits higher correlation with the cycle.

52 The study revealed median monthly figures calculated based on 24-month sliding windows for the sample of European firms in the Moody’s KMV database.

53 This level is in line with those provided in Lopez 2004 and Lee, Lin, and Yang 2011. Stoffberg and van Vuuren 2016 provide evidence that asset correlations in both advanced and EMEs vary from very low levels in good times to substantially positive in bad times. Note in this con-nection that asset correlations are likely to be higher for subsamples of similar borrowers than for broad samples.

TABLE 7.6

Rules of Thumb for the GDP Growth Sensitivity of Credit Risk Parameters


Default rates−0.4 −0.6 −0.8

LGDs−1.5 −2.5 −4.0

Asset correlations (firms)−2.7

Source: Authors, based on studies and sources mentioned in the text.Note: LGDs = losses given default.

bubble and in particular a debt- financed real estate bubble en-tails that much of the collateral backing the lending has suffered a drastic reduction in value, and the market for it has become much less liquid. Hence, LGD rates should be expected to be exceptionally high in such situations. A practical approach could be to use the coefficient from the severe scenario even when the macroeconomic shock is moderate or medium.

Income

As it was done for credit losses, Table 7.4 compares the changes in GDP to changes in preimpairment income. The median GDP sensitivities of income are relatively modest, but the relationship can become highly nonlinear in unfa-vorable circumstances, in line with Figure 7.11, and espe-cially for a substantial portion of poorly performing banks.

Retained earnings

Profits translate into capital directly through the retention of earnings. A bank that makes losses would normally be ex-pected to “retain” all these losses in the form of a reduction in its capital ratio. A bank that makes profits but has low capitalization would be expected to retain all or almost all profits.49 But a bank that is both profitable and well- capitalized would normally pay out a substantial portion of its net income (or extend its balance sheet, when perhaps macroprudential policy is warranted; see IMF 2011b), with-out running down its capital to do so. Such behavior can easily be incorporated into a rule of thumb.50

The expected levels of profit retention by banks under stress are shown in Appendix Table 7.2.1. AE banks are found to pay out about 35 percent in “normal” times and moderate stress times, 20 percent under medium stress, and zero under severe stress. For the EME and LIDC banks, the levels are similar, at around 30 (EME) and 45 (LIDC) respectively in normal times and moderate stress, 17 (EME) and 40 (LIC) respectively under medium stress, and 0 under severe stress.



Box 7.3. How Do IRB Correlations Compare with Empirical Correlations?

As shown in Figure 7.15, internal- ratings- based asset correlations have been calibrated based on conservative rules for “normal” macro- financial conditions. However, due to the nature of the calibration (that is, based on cross- sectional considerations; see main text), correla-tions decrease with an increase in default risk (that is, probability of default), which will lead to underestimation of risk- weighted assets under stress. While there are valid reasons to reduce the procyclicality of regulatory asset correlations, stress testing warrants point- in- time measurement of risks. For medium and severe stress situations, a higher point- in- time correlation level would reflect elevated economy- wide risks, which is often the main concern of stress testers.

Using the benchmarks for the drop in real GDP growth over a three- year horizon (that is, from t = −4 to t = 0) in the run-up of crises (Table 7.4) and the empirical relationship between asset correlations and GDP growth (Appendix Figure 7.2.2), levels of asset correlation for moderate, medium, and severe stress can be estimated. The estimated asset correlations increase substantially with stress, to levels of about 30 percent from 10 percent under “normal” ( through- the- cycle) conditions. However, these estimates are based on extrapolations beyond 16 percent, which warrants a strong caveat. The impact of using different asset correlations on risk- weighted assets is shown in Figure 7.16: the resulting risk weights are similar for moderate stress levels, but become more differentiated under severe conditions, and can reach levels many times greater than normal.

The use of point- in- time asset correlations for stress testing purposes should not result in double counting. However, once losses have materialized the potential for additional losses decreases. Refined calibrations for asset correlations should account for this fact. As an al-ternative, fixed internal- ratings- based correlations based on low probability of default could be used, as suggested by Schmieder, Puhr , and Hasan 2011.

IRB asset correlationsIRB fixed — Schmieder, Puhr, and Hasan 2011Empirical level (Düllmann, Scheicher, and Schmieder 2008)

Normal conditions Moderate stress Medium stress Severe stress0.0

30.0

0.5

10.0

15.0

20.0

25.0

Asse

t cor

rela

tions

for c

orpo

rate

exp

osur

e

Source: Authors, based on IRB formula and Düllmann, Scheicher, and Schmieder 2008.Note: The analysis herein uses the point- in- time probabilities of default and losses given default displayed in Table 7.4 and an effective maturity of 2.5 years. IRB = internal ratings- based.

Figure 7.15 Comparison between IRB Asset Correlation and Empirical Asset Correlations for Corporate Debt (Percent)

IRB asset correlationsIRB fixed — Schmieder, Puhr, and Hasan 2011Empirical level (Düllmann, Scheicher, and Schmieder 2008; this chapter)

Normal Moderate stress Medium stress Severe stress0%

350%

50%

100%

150%

200%

250%

300%

Risk

wei

ght

Source: Authors.Note: The analysis herein uses the point- in- time probabilities of default and losses given default displayed in Table 7.4 and an effective maturity of 2.5 years. IRB = internal ratings- based.

Figure 7.16 Resulting Risk Weights (Percent)



Simulations

The evolution of capital ratios for the AE and EME illustra-tive banks during multiyear periods of stress is shown in Fig-ure 7.17, using the descriptive solvency stress parameters from Appendix Table 7.2.1, except for credit growth, for which a baseline level of 5 percent is used for AEs and 7 per-cent for EMEs.60 For these levels of credit growth, the capi-tal ratios remain roughly unchanged under normal conditions. Year t– 4 shows the initial capital level for banks under both the IRB approach and the StA.

For AE banks under the IRB approach, severe stress will reduce capital levels to 3 percent, that is, below the regula-tory minimum of 8 percent, and medium level stress also reduces capital levels substantially, to about 7 percent from an initial level of 14.7 percent. Moderate stress can be ab-sorbed by banks.

Measured bank capitalization under the StA is generally affected less by stress than that under the IRB approach, largely due to the more substantial changes in RWAs under the IRB approach, especially under severe stress (Table 7.8, which shows RWAs as a proportion of total assets so as to normalize for the effects of balance sheet growth). Variations in RWAs based on point- in- time credit risk parameters (PD, LGD) are very sizable in the absence of mitigation through behavioral adjustments (for example, rebalancing of assets toward highly rated securities).

The leverage ratio (regulatory capital to on- and off- balance-sheet assets) will be hit if income becomes negative (as discussed later in this section), and will rise in the case of a credit contraction. According to the worked examples, se-vere stress in AEs leads to a drop of the leverage ratio by 1.9 percentage points (in the worst year), that is, from 5 per-cent to 3.1 percent (Figure 7.18). For the EME banks, me-dium and severe stress for the EMEs result in a drop of the leverage ratio by 1.4 and 7.9 percentage points, respectively, from an initial level of 7 percent. Hence, “average” banks (as simulated in this case study) are, with the exception of EME banks under severe stress, in a solid position to maintain an adequate leverage ratio.61

Both credit losses and declines in preimpairment income contribute to weakening banks. The typical bank’s net income after loan losses (and therefore retained income) becomes neg-ative under severe stress for the AEs (−25 percent ROC), and under medium (−0.3 percent ROC) and severe stress (−95 per-cent ROC) for the EMEs (Table 7.9). For AE countries, net interest income contributes 59 percent of the decline in earn-ings, followed by net fees and commissions at 21 percent, and

stress situations is needed in order to identify any nonlin-earities (see also Box 7.3, Figures 7.15 and 7.16).

5. WORKED EXAMPLESTo illustrate the use of the descriptive rules of thumb, we simulate the impact of stress on representative stylized banks based in AEs and EMEs, representing weighted average bank characteristics for the sample of banks in Bankscope.54, 55

Bank Characteristics

The illustrative AE bank is assumed to have total on- balance- sheet assets of $100 billion (Table 7.7); customer loans (ex-cluding loans to banks) of $47 billion; $46 billion of securities subject to credit risk; $21 billion of off- balance-sheet assets subject to credit exposure;56 $7 billion in other assets such as fixed assets that are not subject to credit risk; and capital of $6 billion. For the StA, it is assumed that variable external ratings apply to 10 percent of assets that are subject to credit risk (as for nonsovereign bonds, for example), and that their risk weights react accordingly in the stress scenarios. The IRB RWAs are computed under the assumption that RWAs for credit risk amount to 80 percent of total RWAs, and that the risk weights for nonloan assets subject to credit risk are one third of those applied to loans (due to a typically lower default risk, for example, for sovereign exposure and shorter maturi-ties).57 The computed ratio of IRB RWA divided by total as-sets (the so- called RWA density) is in line with the observed average RWA density for large international IRB banks.

The corresponding bank based in an EME is assumed to have the same total assets, of which 54 percent constitutes customer loans, with capital amounting to $8.7 billion. The elevated credit risk (that is, higher PDs and LGDs) of EME banks compared to that of AE banks results in relatively high capital weights under the IRB approach, and therefore a lower capital ratio.58, 59

54 Using the satellite model rules of thumb with the benchmark growth rates would give very similar results.

55 For all characteristics but income and credit growth, the parameters used in the worked example correspond to realistic postcrisis character-istics, while all figures related to income and credit growth represent long- term averages in order to mimic banks’ structural characteristics.

56 This includes credit lines, credit guarantees and alike. We assume a credit conversion factor of 50 percent, that is, $21 billion in off- balance sheet credit exposure is equivalent to $10.5 billion on- balance sheet.

57 Further, it is assumed that 40 percent of the customer loans are large corporate, 20 percent small and medium- sized enterprises, and 40 per-cent retail and that the effective maturity is 2.5 years. The PDs and LGDs for corporate loans are assumed to be as reported in Table 7.3 under normal conditions (the LGD is set to 30 percent). A scaling factor of 1.5 is applied to both parameters for percent small and medium- sized enterprises and a scaling factor of 0.75 is applied to retail exposure. For simplicity, other RWAs components are assumed to be proportional to the credit- related RWAs components.

58 A risky asset could require more than 8 percent minimum capital, im-plying a risk weight over 100 percent.

59 The fact that the EME’s IRB- based capital ratio (8.1 percent) under normal conditions is well below the EME’s StA capital ratio (14.0 per-cent) suggests that EME banks should not normally expect an increase of capital ratios when they move to the IRB approach.

60 For EMEs, the growth rates for the stress scenarios are adjusted propor-tionally (that is, using growth rates of 7 percent, 5 percent, 3 percent, and 1 percent under normal conditions, moderate stress, medium stress, and severe stress, respectively), while the data from Table 7.4 are used for the AEs except for a growth level of 4.6 percent under “normal” condi-tions. Details of the worked example are available from the authors.

61 This conclusion would not hold for a bank that grows its portfolio very rapidly. During the last 15 years, average credit growth rates were about 10 percent in AEs and slightly above 20 percent in EMEs.




0

181. IRB Approach, Advanced Economies

–4 –3 –2 –1 0 31 2

–4

163. IRB Approach, Emerging Market Economies

–4 –3 –2 –1 0 31 2

Years around crisis

Years around crisis

0

182. StA, Advanced Economies

–4 –3 –2 –1 0 31 2

–4

164. StA, Emerging Market Economies

–4 –3 –2 –1 0 31 2

Years around crisis

Years around crisis

2

4

6

8

10

12

14

16

–202468

101214

2

4

6

8

10

12

14

16

–202468

101214

Source: Authors, based on Bankscope data.Note: IRB = internal-ratings- based; StA = standardized approach.

Figure 7.17 Evolution of Capital Ratios during Stress Periods (Percent)

TABLE 7.7

Features of Banks Used in the Worked Examples (Billions of US dollars)Example Bank AEs EMEs

Assets 100.0 100.0 of which: Customer loans 47.0 54.0 Other on-balance-sheet assets subject to credit risk 46.0 39.0 Assets not subject to credit risk 7.0 7.0Off-balance-sheet assets subject to credit risk 21.0 24.0Total regulatory capital 6.0 8.7Leverage ratio (capital/[on + off-balance-sheet assets]) 5.0 7.1StA RWAs (billions of US dollars)1 64.3 62.0StA capital ratio (percent)2 9.3 14.0Implied IRB RWAs (billions of US dollars)3 40.9 107.4 Implied IRB capital ratio (percent)2 14.7 8.1Preimpairment income (ROC, percent, long-term average) 12.0 20.0Net income (billions of US dollars, long-term average) 0.4 1.1ROC (percent, long-term average) 6.7 9.3

Source: Authors, based on Bankscope data.Note: AEs = advanced economies; EMEs = emerging market economies; IRB = internal-ratings-based; ROC = return on capital; RWAs = risk-weighted assets; StA = standardized approach.1AE: weighted average for banks with total assets less than $100 billion. EME: weighted average of all banks.2Capital divided by respective RWAs.3Estimated based on the sensitivities presented in this study, assuming that 20 percent of RWAs are accounted for by other risk types (such as market risk and operational risk).



other income at 20 percent (trading and fair value contributes half of this). For EME banks, the corresponding contribu-tions are 74 percent, 18 percent, and 9 percent. In order to simulate the income of a specific bank under stress, the sensi-tivities given in Table 7.5 could be used together with the rela-tive contribution of the earning sources for that specific bank to adjust the trajectories of net overall income.

When compared to evidence from the Bankscope data-set, the predicted evolution of capitalization of an AE bank under severe stress based on the StA roughly matches the evolution of the capital ratio of the 25th percentile of those banks that experienced severe stress conditions during the last 15 years (Figure 7.19, which shows the evolution of ac-tual, historical capitalization for various quantiles, and the


0

6

–4 –3 –2 –1

Years around crisis

0 1 2 3

1

5

3

4

2

–4 –3 –2 –1

Years around crisis

0 1 2 3–2

–1

0

1

2

3

4

5

6

7

81. Leverage Ratio, Advanced Economies 2. Leverage Ratio, Emerging Market Economies


Figure 7.18 Evolution of Leverage Ratios during Stress Periods (Percent)

TABLE 7.8

Simulated Evolution of RWAs Relative to Total Assets during Stress Periods(Percent)

t = −3 t = −2 t = −1 t = 0 t = 1 t = 2 t = 3

RWA StA

AEsNormal 64 64 64 64 64 64 64Moderate 64 64 65 71 64 64 64Medium 64 65 73 79 64 64 64Severe 65 73 85 88 65 65 65

EMEsNormal 63 63 63 63 63 63 63Moderate 63 63 64 69 63 63 63Medium 63 64 71 77 63 63 63Severe 64 71 80 87 64 64 64

RWA IRB Approach

AEsNormal 41 41 41 41 41 41 41Moderate 41 41 41 65 41 41 41Medium 41 41 67 87 66 42 42Severe 41 67 93 165 90 69 44

EMEs

Normal 108 109 109 110 110 110 111Moderate 108 109 111 148 111 111 112Medium 108 111 153 202 153 115 115Severe 110 153 207 340 230 177 134

Source: Authors.Note: AEs = advanced economies; EMEs = emerging market economies; IRB = internal-ratings-based; RWAs = risk-weighted assets; StA = standardized approach.



levels predicted by the descriptive rules of thumb).62 The capitalization prediction based on the IRB approach is close to the 10th percentile of actual results. The comparability of outcomes is reassuring, suggesting that the calibration is ap-propriate, and data by Dagher and others (2016), who at the sufficiency of capital during banking crises paints a broadly similar picture.63

62 This simulation assumes that the bank used a capital regime comparable to the StA in the past, which is by and large valid (Basel I RWAs were not sensitive to changes in risk, but accounted for volume and asset class mix). There is insufficient past evidence for banks using the IRB approach or for EM and LIC banks. European and Japanese banks have reported capital ratios based on IRB parameters since 2008.

63 It should be noted that actual capital ratios will reflect also managerial action by banks, such as capital raisings in capital markets, selling of legal entities, on so on, which are not captured by the simulated capital ratio. Hence, it is in line with expectations that the median of the actual capital ratio is above the simulated trajectory for the capital ratios.

Actual top quartile Actual median Actual lowest quartile Actual lowest decileActual 5th percentile Predicted StA Predicted IRB

–4

14

Years around crisis

StA-based prediction

IRB-based prediction

–3 –2 –1 0

–2

0

2

4

6

8

10

12

Source: Authors, based on Bankscope data (2,050 observations).Note: AE = advanced economy; IRB = internal- ratings- based; StA = standardized approach.

Figure 7.19 Evolution of Capital Ratios: Actual vs. Predicted for an AE Bank (Percent)

As another means of comparison, the rules for the satel-lite models were applied in the context of recent stress tests run for EU and US banks. Again the results were reassuring (Box 7.4).

Rules of Thumb, the Regulatory Regime, and Procyclicality

The examples illustrate that a capital ratio projected under the StA tends to react less quickly (in either direction) than one projected using the IRB approach, and therefore needs to be interpreted differently. Even a relatively modest change in StA- measured capitalization may be economically signifi-cant. By the same token, IRB capital ratios require ample buffers to withstand a given level of stress (especially if based on point- in- time PDs and LGDs)—as high IRB capital ratios in good times can give a false sense of security, while they are

TABLE 7.9

Simulated Evolution of Net Income during Stress Periods (Percent ROC)

t = −3 t = −2 t = −1 t = 0 t = 1 t = 2 t = 3

AEsModerate 7.0 6.9 6.5 3.0 6.6 6.1 6.2Medium 7.0 6.4 2.1 −0.3 1.2 5.1 5.1Severe 7.0 5.8 −2.4 −25.1 −6.1 −1.7 −0.4

EMEsModerate 9.7 9.8 9.8 2.5 10.0 8.9 9.6Medium 9.7 9.8 2.5 −5.8 −0.9 8.0 5.7Severe 9.8 9.5 −5.5 −95.1 −290.5 … 1 … 1

Source: Authors.Note: AEs = advanced economies; EMEs = emerging market economies; ROC = return on capital.1Undefined due to negative capital.



higher initial risk weights than AE banks, and there-fore the sensitivity of RWA to cyclical fluctuations is lower.

A common concern is that the regulatory regime may im-pose capital requirements that are excessively procyclical; capital requirements would increase sharply in difficult times if they impose use of point- in- time parameters. Such procyclicality can be costly and even destabilizing if it pro-vokes a credit crunch that further weakens economic perfor-mance. One could replace point- in- time- based RWAs by RWAs based on a moving average of, say, for a five- year pe-riod, which can be considered a through- the- cycle estimate for regulatory purposes. The counterargument is that, for purposes of stability analysis, point- in- time estimates are what matter, as the through- the- cycle estimates are not rele-vant to assessing whether a bank will survive through the strains to which it is subject at the worst phases of the cycle.64

The rules of thumb allow one to compare the effects of the two approaches (Figure 7.20). A through- the- cycle ap-proach results in a slower decline in calculated capital ratios than does a point- in- time approach over the two years lead-ing to a crisis. Under through- the- cycle, a substantial drop comes at the time of the crisis or thereafter. Hence, differ-ences in capital ratios end close to the starting level (as one would expect, assuming that neither approach is biased across the cycles as a whole). The point- in- time approach

conservative in bad times. Hence, in order to interpret a sol-vency measure and in particular the results of a stress test simulation, one needs to understand at which stress level a bank currently finds itself, and the regulatory regime.

The stylized example allows one to compute the impact of a drop of GDP growth by 1 percentage point on capital ra-tios under different stress levels, using the historical average changes of GDP growth levels corresponding to these stress levels (as displayed in Table 7.4 and Figures 7.17 and 7.19):

• For the AE bank, a 1-percentage-point drop of GDP growth would lead to a reduction in capitalization by 0.4 to 0.5 percentage points for an StA bank, and about 1.6 to 2.3 percentage points for an IRB bank, depending on the severity of the shock. For the cu-mulative deviation from the GDP growth trend, the coefficients should to be divided by two. While this is the general trend based on median rules, the effect could be nonlinear and not strictly monotonic be-cause several factors affect the capital ratio: a stress that results in a modest rise in credit loss but a sharp deceleration in credit growth can leave the capital ratio more or less unchanged, for example.

• For the EME bank, the corresponding loss of capi-talization would be about 0.5 to 1.2 under the StA per unit drop of real GDP growth, and 0.6 to 0.7 percentage points under the IRB approach. (For the cumulative deviation from the GDP growth trend, the coefficients should be divided by three.) The coefficient under the IRB approach for EME banks is relatively low because EME banks exhibit

Box 7.4. Rules of Thumb Applied to Previous Stress Tests

The satellite model rules of thumb are applied to the scenarios used in the stress tests run by the European Banking Authority (EBA 2011) and the US authorities (Board of Governors of the Federal Reserve System 2012), both of which are similar to the ones applied more re-cently. The macroeconomic scenario for the EU simulated a cumulative deviation from baseline growth by about 4 percentage points for a horizon of two years (2011–12). The scenario for the United States simulated a cumulative deviation during three years (2012–14) by 5 per-centage points from the baseline, including a dip by 6 percentage points within one year (as observed in 2009), with a subsequent recov-ery. The European stress tests included 90 large banks, and the United States tested the largest 18 banking holding companies. For the computations in this box, the worked example with the average AE bank will be assumed to be representative for the average of both Eu-ropean and US banks.

The outcome of the EBA stress test— an average drop of capital ratios by around 2 percentage points (without considering capital increase)—can be approximated both by the “cumulative deviation from GDP trend” rule and by the “change in GDP growth” rule (Table 7.4). The latter rule is probably more useful for the scenario at hand, which did not simulate a sharp drop for one year, but rather a sustained lower level of real GDP growth of around zero (compared to 2 percent under the baseline). Using the average AE bank from Table 7.7, the IRB rule of thumb (applicable because most large European banks use the IRB approach) linking GDP growth to changes in capital ratios suggests a drop of capital ratios by 3 to 4 percentage points based on the GDP change rule (a 2-percentage-point drop in real GDP growth times sensitivity of 1.5–2 percentage points per unit of GDP), while the cumulative deviation rule would suggest a drop of capital ratios by about 3.2 percentage points (a 4-percentage-point drop times sensitivity of about 0.8 percentage points per unit of GDP (see next section). Both scenarios would fall into the “moderate” category (Table 7.4). This outcome is reasonable, while reflecting that the rules of thumb are intentionally conservative.

For the test run for the US banks, the average drop of capital ratios during the dip year (in which GDP growth was simulated to drop from +2 percent to −4 percent) was estimated at around 4.3 percentage points. The StA rule of thumb (applicable because US banks cur-rently apply Basel I) for the change rule (corresponding to a medium/severe stress level) would suggest a drop by about 3.9 percentage points (6-percentage-point drop times sensitivity of about 0.65 percentage points per unit of GDP), which is again relatively close to the actual estimate.

64 Similar considerations apply for asset correlations, as discussed previ-ously in this chapter.



It should be noted that the policy considerations warrant two caveats:

• The examples simulate average behavior of banks in terms of credit growth (that banks do not deleverage more as suggested by the rules of thumb), assuming that banks do not change the structure of their bal-ance sheet (for example, by replacing assets with higher risk weights by assets with lower risk weights), and neglecting asset sales (for example, by sale of as-sets or disposals of legal entities) as well as capital raising beyond an endogenous amount of retained earnings. For multiyear periods, these are strong as-sumptions, but with the advantage of yielding a clear- cut benchmark of what would happen without major changes.

• The simulation uses median risk parameters for banks that experienced severe stress. While median credit loss rates under severe stress are close to 3 per-cent for AE banks, they were much higher for the below- average bank (Table 7.4 and Figure 7.18). The same applies to income, where banks with more volatile sources of income can experience income levels that are substantially more unfavorable than those of the median bank (Figure 7.10). The same applies to the other severity levels, but in those cases the range of parameters is lower (given that they are bound by the severe case).

6. CONCLUSIONA variety of evidence is presented on the “average” pattern of behavior of financial aggregates relevant to solvency stress testing banks based in EMEs and AEs, and, with some limi-tations, also for larger LIDC banks. Table 7.10 provides an overview of some main results.

gives an earlier warning of strain, and the through- the- cycle approach perhaps gives more “breathing space” during which capital can be bolstered, and maturing assets could be replaced by assets with lower risk weights.

Policymakers might well ask what level of the capital ra-tio is needed, such that a (typical) bank stays above the regu-latory minimum (8 percent total capital ratio) under stress (see also Dagher and others 2016). To answer this question, the required capital levels are simulated by changing the capital ratios (via the numerator) to determine the initial level that would result in a ratio of about 8 percent at time t = 0 when strain is greatest. An average AE bank under the IRB approach would need to achieve a capital ratio of around 30 percent in normal times in order to remain capi-talized at or above 8 percent at time zero under severe stress and assuming a point- in- time capital regime; it would need about 17 percent to be prepared for medium stress. Using through- the- cycle risk weights, the corresponding capital ra-tios would be 20 percent for severe stress and 14 percent for the median scenario. For the average AE bank under the StA, a capital ratio of 10 percent would be needed to cope with medium stress without falling below the regulatory floor, and 12.5 percent to cope with extreme stress.

For an EME bank under the StA, a capital level of 12 per-cent would be sufficient to cope easily with medium level shocks, and a capital level of around 22 percent would be sufficient to cope with severe stress.

Banks would need more capital if they had relatively volatile preimpairment earnings. For example, an AE bank from the worst decile of preimpairment earnings would need a capital ratio of 22 percent using the IRB approach to cope with medium stress (Figure 7.10). An EME bank from the worst decile, using StA, would need a capital ratio of 17 per-cent to deal easily with medium stress, and 25 percent to cope with severe stress.

AE_PIT EME_PIT AE_TTC EME_TTC

Years around crisis

–2

16

–3 –2 –1 0 1 2 3

Years around crisis

0

2

4

6

8

10

12

14

Source: Authors.Note: AE = advanced economy; EME = emerging market economy; PIT = point in time; RWAs = risk- weighted assets; TTC = through the cycle.

Figure 7.20 Capital Ratios with Point- in- Time vs. Through- the- Cycle RWAs (Percent)



While the study has found general patterns, country- specific and/or bank- specific circumstances may differ widely from the average. Hence, the rules of thumb elabo-rated in this study serve as broad guidelines, particularly to understand benchmarks for worst- case scenarios, but do not fully substitute for detailed analysis when that is possible. The rules of thumb with explicit focus on macro- financial linkages cover only some of the main macroeconomic risk factors that may affect a banking system, namely those cap-tured by GDP. It would be worthwhile to investigate whether analogous simple rules can be formulated that link specific elements of banks’ balance sheets and profitability to such other sources of vulnerability.65

The rules of thumb can be used to compute minimum lev-els of capitalization needed to withstand shocks of different severities— even those far from a country’s historical experi-ence. Also, the regulatory approach used by banks matters: whether a bank adopts an IRB approach to estimating RWAs or relies on an StA is shown to make a substantial difference to the magnitude and also the timing of when the effects of shocks are recognized, provided that banks’ risk models re-flect changes in risk on a timely basis.66 Thus, the results are relevant to the design of (countercyclical) capital buffers (for example, Drehmann and others 2010) and broader discus-sions on the design of regulation (for example, Haldane 2012, 2013). The results also echo the call for (much) longer samples to be used in the calibration of models used for RWA compu-tation (for example, BCBS 2013 and BIS 2013).

Typical levels of credit loss rates, preimpairment income, and credit growth were estimated under moderate stress (a one- in- 10- to- 15-year shock), medium stress ( worst- in- 20-years shock), severe stress (a 1- in- 40-year shock), and extreme stress (1- in- 100-year shock). All three variables react in nonlinear fashion to the severity of stress, which means that effects un-der severe conditions are many times the effects under moder-ate conditions. Also, a substantial “tail” of poorly performing banks is likely to be much more affected than the median bank.

Comparing AEs on the one hand and EMEs and LIDCs on the other, loss levels are found to be substantially higher in the latter two groups, compensated for by higher returns. It was found that 1- in- 20-year stress loss levels usually lead banks to report some net losses, especially in EMEs, and thereby lose some capitalization (1 to 3 percentage points if they are under Basel I or the Basel II StA), but only a macro-economic crisis approaching severe intensity would normally bring down typical well- capitalized banks (unless there are other issues related to confidence and financialsector- generated sources of strain).

Further evidence is presented on macro- financial linkages, and specifically on defining rules of thumb of how a change in GDP growth triggers credit losses, income, and credit growth effects under different levels of stress. While such rough satel-lite models are more complex than the descriptive solvency rules, they allow the development of scenarios based on an explicit story. As such, the rules make allowance for national circumstances, such as the expected severity of shocks.

65 Relevant macroeconomic variables could include (1) interest move-ments, including an overall shift in rates and a steepening or flattening of the yield curve (effects are likely to depend crucially on how fre-quently rates on various assets and liabilities adjust); (2) inflation and especially unexpected movements in the inflation rate (a rapid decelera-tion could strain borrowers’ ability to repay); (3) exchange rate move-ment, especially where a large proportion of loans are denominated in foreign currency; and (4) shocks affecting sectoral concentration of ex-posures or certain business lines.

66 Note that the simulation does not take into account recent policy pro-posals by the Basel Committee on Banking Supervision in this context (for example, BCBS 2017), but the broad pattern remains the same.

TABLE 7.10

Overview of Main Rules of Thumb(Percent)

Normal Medium Stress Severe Stress

AE bankAnnual credit loss rate 0.3 1.1 2.4Preimpairment ROC 11.9 11.4 8.3Credit growth 7.2 1.3 −3.8Asset correlation 10.4 21.8 30.1

EME bankAnnual credit loss rate 1.0 3.4 7.4Preimpairment ROC 18.9 18.6 13.4Credit growth rate 18.9 3.2 −8.3

Source: Authors; and evidence in chapter.Note: AE = advanced economy; EME = emerging market economy; ROC = return on capital.


Appendix 7.1.Data Summary


Overview of Bankscope DataType of Country

Number of Countries

Number of Banks

Assets (Trillions of US dollars)

Original raw dataAE 33 13,271 120.4EME 109 3,126 28.2LIDC 59 543 1.2Total 201 16,940 149.9

Cleaned, final dataAE 32 9,372 68.8EME 90 1,131 10.4LIDC 47 206 0.3Total 169 10,709 79.5



Appendix Figure 7.1.1 Overview of Raw Bankscope Sample Size by Year (Number of banks)

0

18,000

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

1996 97 98 99 2000 01 02 03 04 05 06 07 08 09 10 11




Overview of Top 10 Countries by Category for Cleaned Bankscope DataCountry (ISO code) Number of Banks Total Assets

(In billions of US dollars)

Advanced economiesUS 5,930 11,447DE 1,513 6,428JP 497 10,030IT 398 4,043AT 194 967CH 165 330FR 137 10,190ES 89 3,692GB 78 9,026DK 71 1,050

Emerging market economiesRU 96 773BR 73 2,149AR 49 118IN 39 438MY 30 496UA 29 44ID 28 78DO 26 19HR 25 57TW 24 693

Low-income countriesKE 19 12NG 15 78BD 13 28TZ 11 2ZM 9 3ET 8 9UG 8 1AO 7 15KH 6 2YE 6 0

Source: Authors, based on Bankscope data.Note: This table uses International Organization for Standardization (ISO) abbreviations for country names.


Overview of Bankscope Data, by Stress LevelTotal Moderate Stress Medium Stress Severe/Extreme Stress

Number of banksAE 9,372 1,952 4,402 3,018EME 1,131 305 491 335LIDC 206 66 77 64

Assets in billions of US dollarsAE 68,767 14,666 35,977 18,124EME 10,427 2,734 5,102 2,592LIDC 347 129 80 138




Appendix 7.2.Supplementary Evidence


Evolution of Solvency Parameters around Crisis DatesYears around Crisis

−3 −2 −1 0 1 2 3

Credit loss rates/total customer loansAE “Normal” 0.3 0.3 0.3 0.3 0.3 0.3 0.3AE Moderate 0.2 0.2 0.4 0.8 0.4 0.3 0.2AE Medium 0.3 0.4 0.7 1.5 0.7 0.5 0.5AE Severe 0.3 0.5 1.2 4.0 1.3 0.7 0.5EME “Normal” 1.0 1.0 1.0 1.0 1.0 1.0 1.0EME Moderate 0.7 0.8 1.2 2.5 1.2 0.9 0.7EME Medium 1.1 1.3 2.3 5.2 2.0 1.2 0.9EME Severe 2.1 2.4 4.1 15.6 3.5 1.8 1.1LIDC “Normal” 1.4 1.4 1.4 1.4 1.4 1.4 1.4LIDC Moderate 1.3 1.2 1.6 3.2 1.2 1.0 0.8LIDC Medium 1.4 0.9 2.1 6.4 1.9 1.2 1.0LIDC Severe 2.8 3.7 3.0 15.3 4.1 1.3 1.7

Credit growth (adjusted for losses)AE “Normal” 7.2 7.2 7.2 7.2 7.2 7.2 7.2AE Moderate 7.5 7.5 6.0 3.5 3.2 3.8 4.7AE Medium 7.0 6.3 4.2 1.3 1.2 2.5 4.0AE Severe 11.0 8.9 3.6 −3.8 −4.3 −0.3 2.6EME “Normal” 22.7 22.7 22.7 22.7 22.7 22.7 22.7EME Moderate 19.8 21.8 19.3 11.8 14.4 18.3 22.7EME Medium 29.8 26.9 16.1 7.8 13.8 21.8 24.3EME Severe 23.8 20.1 11.6 1.3 15.5 24.9 23.6LIDC “Normal” 20.5 20.5 20.5 20.5 20.5 20.5 20.5LIDC Moderate 13.3 30.0 36.2 22.0 21.8 28.8 21.4LIDC Medium 31.7 17.1 23.9 12.3 14.9 26.2 23.4LIDC Severe 19.8 20.5 24.5 13.0 15.5 30.3 20.7

Preimpairment income/capitalAE “Normal” 11.9 11.9 11.9 11.9 11.9 11.9 11.9AE Moderate 13.9 13.8 13.2 13.5 13.1 12.5 12.6AE Medium 14.2 13.4 12.7 12.5 11.4 11.2 11.2AE Severe 14.4 12.9 10.5 8.0 8.3 8.9 9.8EME “Normal” 18.9 18.9 18.9 18.9 18.9 18.9 18.9EME Moderate 22.1 21.8 21.0 24.2 22.0 20.7 21.4EME Medium 22.9 21.2 23.0 23.7 18.6 18.7 16.2EME Severe 17.6 21.9 23.2 26.2 14.4 13.4 17.9LIDC “Normal” 25.0 25.0 25.0 25.0 25.0 25.0 25.0LIDC Moderate 10.2 13.2 18.7 24.7 30.2 32.8 36.3LIDC Medium 63.5 47.6 30.4 30.0 22.3 23.7 21.5LIDC Severe 15.4 16.5 11.8 41.1 29.2 15.5 25.1

Dividend payout/net incomeAE “Normal” 33.7 33.7 33.7 33.7 33.7 33.7 33.7AE Moderate 41.6 42.9 40.0 34.9 34.0 33.3 38.5AE Medium 37.4 37.1 35.1 20.0 22.5 27.9 25.0AE Severe 23.9 23.2 0.0 0.0 0.0 3.6 17.0EME “Normal” 28.8 28.8 28.8 28.8 28.8 28.8 28.8EME Moderate 26.8 34.7 25.8 31.3 27.5 21.6 28.1EME Medium 21.3 22.3 22.6 17.1 25.3 24.8 25.5EME Severe 24.4 22.4 13.1 0.0 11.7 23.7 23.7LIDC "Normal" 44.3 44.3 44.3 44.3 44.3 44.3 44.3LIDC Moderate 37.6 46.9 41.2 44.9 37.2 49.4 53.5LIDC Medium 46.4 48.4 32.6 40.8 47.2 52.3 39.2LIDC Severe 44.8 33.3 37.1 0.0 42.9 35.0 38.5

(continued)



Tax payments/pretax net incomeAE “Normal” 28.2 28.2 28.2 28.2 28.2 28.2 28.2AE Moderate 28.0 27.2 26.1 25.2 26.2 27.2 28.4AE Medium 30.5 29.3 28.4 23.4 27.1 29.4 32.3AE Severe 30.2 29.3 26.7 15.7 18.9 24.3 26.6EME “Normal” 20.6 20.6 20.6 20.6 20.6 20.6 20.6EME Moderate 23.1 21.9 22.0 22.2 21.7 22.1 22.9EME Medium 20.6 21.5 20.0 17.9 18.7 18.1 21.0EME Severe 19.6 21.0 18.6 9.9 12.5 14.7 17.8LIDC “Normal” 27.0 27.0 27.0 27.0 27.0 27.0 27.0LIDC Moderate 27.9 33.0 33.4 31.3 33.6 31.5 31.3LIDC Medium 29.0 30.6 30.1 27.8 30.4 30.8 29.9LIDC Severe 33.0 32.9 32.5 23.7 20.9 28.2 27.0






Components of Preimpairment Income and Expenses (Percent of capital)


Years around crisis

−3 −2 −1 0 1 2 3

Normal conditionsNet preimpairment income 13.1 13.1 13.1 13.1 13.1 13.1 13.1Net interest income 32.8 32.8 32.8 32.8 32.8 32.8 32.8Fees 1.8 1.8 1.8 1.8 1.8 1.8 1.8Other operating income 4.8 4.8 4.8 4.8 4.8 4.8 4.8Operating expenses −26.6 −26.6 −26.6 −26.6 −26.6 −26.6 −26.6

Moderate stressNet preimpairment income 13.9 13.8 13.2 13.5 13.1 12.5 12.6Net interest income 33.7 33.2 32.9 32.9 32.4 31.9 32.5Fees 5.1 5.1 2.0 0.7 0.5 0.7 1.6Other operating income 5.2 4.9 4.7 4.7 4.5 4.7 4.8Operating expenses −26.4 −26.1 −26.0 −26.0 −26.1 −26.0 −25.2

Medium stressNet preimpairment income 14.2 13.4 12.7 12.5 11.4 11.2 11.2Net interest income 34.1 33.8 33.2 33.1 32.3 31.8 32.1Fees 5.1 5.3 1.2 0.4 0.4 0.8 3.3Other operating income 5.4 5.3 5.0 4.6 4.3 4.1 4.3Operating expenses −27.2 −27.5 −27.6 −27.5 −27.4 −27.4 −27.7

Severe stressNet preimpairment income 14.4 12.9 10.5 8.0 8.3 8.9 9.8Net interest income 34.7 33.7 32.5 35.6 34.1 32.5 32.7Fees 4.4 4.6 0.6 0.2 0.1 0.2 0.8Other operating income 5.0 4.6 4.5 5.6 5.5 4.9 5.2Operating expenses −26.8 −27.2 −29.0 −35.1 −32.7 −31.0 −29.6

2. Emerging Market Countries

Years around crisis

−3 −2 −1 0 1 2 3

Normal conditionsNet preimpairment income 19.6 19.6 19.6 19.6 19.6 19.6 19.6Net interest income 30.3 30.3 30.3 30.3 30.3 30.3 30.3Fees 7.3 7.3 7.3 7.3 7.3 7.3 7.3Other operating income 10.4 10.4 10.4 10.4 10.4 10.4 10.4Operating expenses −24.7 −24.7 −24.7 −24.7 −24.7 −24.7 −24.7

Moderate stressNet preimpairment income 22.1 21.8 21.0 24.2 22.0 20.7 21.4Net interest income 30.5 29.1 30.4 31.7 32.1 32.3 33.2Fees 8.3 6.7 6.8 6.4 6.7 6.4 5.8Other operating income 11.9 9.1 8.5 9.3 9.7 10.4 9.0Operating expenses −27.6 −25.5 −23.4 −22.2 −22.1 −24.2 −22.7

Medium stressNet preimpairment income 22.9 21.2 23.0 23.7 18.6 18.7 16.2Net interest income 31.5 30.9 34.4 32.5 28.8 27.8 26.8Fees 9.8 8.8 8.0 7.9 8.1 8.2 8.5Other operating income 12.9 11.6 10.5 10.2 10.7 10.9 10.7Operating expenses −27.7 −26.5 −25.8 −24.9 −25.4 −23.0 −23.6

Severe stressNet preimpairment income 17.6 21.9 23.2 26.2 14.4 13.4 17.9Net interest income 28.7 34.3 37.1 32.9 24.2 27.4 27.6Fees 8.9 7.8 6.9 8.0 6.9 6.4 7.3Other operating income 11.5 11.7 11.6 12.2 10.2 10.5 9.4Operating expenses −33.1 −30.6 −32.4 −33.7 −26.9 −25.3 −23.1

Source: Authors, based on Bankscope data.Note: The figures are medians, and therefore they do not necessarily add up.



Source: Authors, based on Düllmann, Scheicher, and Schmieder 2008.

Appendix Figure 7.2.1 Asset Correlations (Percent)

0

18

Jan. 03

Asse

t cor

rela

tion

(per

cent

)

2

4

6

8

10

12

14

16

Jan. 1997 Jan. 98 Jan. 99 Jan. 2000 Jan. 01 Jan. 02

Minimum: 4.3Mean: 10.5

Median: 11.2Maximum: 16.2

Source: Authors, based on Düllmann, Scheicher, and Schmieder 2008.

Appendix Figure 7.2.2 Asset Correlations and GDP Growth Rates (Percent)

0

18

Real GDP growth

Asse

t cor

rela

tion

2

4

6

8

10

12

14

16

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Asset correlation = –2.66RGDP gr. + 16.76R 2 = 0.44



/ How- Well- Do- Aggregate- Bank-Ratios-Identify-Banking -Problems-21464.

Dagher, Jihad, Giovanni Dell’Ariccia, Luc Laeven, Lev Ratnovski, and Hui Tong. 2016. “Benefits and Costs of Bank Capital.” IMF Staff Discussion Note 16/04, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications / Staff- Discussion- Notes/Issues/2016/12/31/ Benefits- and-Costs -of-Bank-Capital-43710.

Drehmann, Mathias. 2009. “Macroeconomic Stress Testing Banks: A Survey of Methodologies.” Stress Testing the Banking System: Methodologies and Applications, edited by Mario Qua-gliariello. Cambridge: Cambridge University Press.

Drehmann, Mathias, Claudio Borio, Leonardo Gambacorta, Ga-briel Jiminez, and Carlos Trucharte. 2010. “Countercyclical Capital Buffers: Exploring Options.” BIS Working Papers 317, Bank for International Settlements, Basel, Switzerland. https://www.bis.org/publ/work317.htm.

Düllmann, Klaus, Martin Scheicher, and Christian Schmieder. 2008, “Asset Correlations and Credit Portfolio Risk— An Em-pirical Analysis.” Journal of Credit Risk 4 (2): 37–62.

European Banking Authority. 2011. “ EU- Wide Stress Testing 2011.” London. http:// stress- test.eba.europa.eu/web/guest/ risk - analysis- and-data/eu-wide-stress-testing/2011.

Foglia, Antonella. 2009. “Stress Testing Credit Risk: A Survey of Authorities’ Approaches.” International Journal of Central Banking 5 (3): 9–45.

Giesecke, Kai, Francis A. Longstaff, Stephen Schaefer, and Ilya Strebulaev. 2011. “Corporate Bond Default Risk: A 150-Year Perspective.” Journal of Financial Economics 102 (2): 233–250.

Gigerenzer, Gerd, Ralph Hertwig, and Thorsten Pachur. 2011. Heuristics: The Foundations of Adaptive Behavior. Oxford: Ox-ford University Press.

Haldane, Andrew G. 2012. “The Dog and the Frisbee.” Bank of England speech at the Federal Reserve Bank of Kansas City’s 366th economic policy symposium, The Changing Policy Land-scape, Jackson Hole, Wyoming, August 31. https://www .bankofengland.co.uk/paper/2012/the-dog-and-the-frisbee.

———. 2013. “Constraining Discretion in Bank Regulation.” Paper presented at the Federal Reserve Bank of Atlanta Confer-ence, “Maintaining Financial Stability: Holding a Tiger by the Tail(s),” Georgia, April 9. https://www.bankofengland.co.uk /paper/2013/constraining-discretion-in-bank-regulation.

Hardy, Daniel C., and Ceyla Pazarbaşioğlu. 1999. “Determinants and Leading Indicators of Banking Crises: Further Evidence.” IMF Staff Papers 46 (3): 247–258. https://www.imf.org/exter nal/pubs/ft/staffp/1999/09-99/hardy.htm.

Hardy, Daniel C., and Christian Schmieder. 2013. “Rules of Thumb for Bank Solvency Stress Testing.” IMF Working Paper 13/232, International Monetary Fund, Washington, DC.https://www.imf.org/en/Publications/WP/Issues/2016/12 /31/ Rules- of- Thumb- for-Bank-Solvency-Stress-Testing-41047.

International Monetary Fund (IMF). 2010. “Resolution of Cross- Border Banks— A Proposed Framework for Enhanced Coordi-nation.” IMF Policy Paper, Washington, DC. https://www.imf .org/en/Publicat ions/ Policy- Papers/Issues/2016/12/31 / Resolution- of- Cross- Border- Banks- A- Proposed-Framework -for-Enhanced-Coordination-PP4462.

———. 2011a. Global Financial Stability Report, September 2011—Grappling with Crisis Legacies, Chapter 1. Washington, DC, September. https://www.imf.org/en/Publications/GFSR /Issues/2016/12/31/ Globa l- Financia l- Stabi l ity- Report - September- 2011-Grappling-with-Crisis-Legacies-24745.

REFERENCESAlfaro, Rodrigo, and Matthias Drehmann. 2009. “Macro Stress

Tests and Crises: What Can We Learn?” BIS Quarterly Review December, Bank for International Settlements, Basel, Switzer-land. https://www.bis.org/publ/qtrpdf/r_qt0912e.htm.

Araten, Michael, Michael Jakobs Jr., and Varshney Peeyush. 2004. “Measuring LGD on Commercial Loans: An 18-Year Internal Study.” The RMA Journal (May): 28–35.

Bank of International Settlements (BIS). 2013. “The Road to a More Resilient Banking Sector.” 83rd Annual Report 2012/13. Basel: Bank for International Settlements. https://www.bis.org /publ/arpdf/ar2013e.htm.

Basel Committee on Banking Supervision (BCBS). 2006. Results of the Fifth Quantitative Impact Study (QIS 5). Basel: Bank for In-ternational Settlements. https://www.bis.org/bcbs/qis/qis5.htm.

———. 2011. Basel III: A Global Regulatory Framework for More Resilient Banks and Banking Systems. Revised version, June. Basel: Bank for International Settlements. https://www.bis.org /publ/bcbs189.htm.

———. 2013. Regulatory Consistency Assessment Programme (RCAP)—Analysis of Risk- Weighted Assets for Credit Risk in the Banking Book. Basel: Bank for International Settlements. https://www.bis.org/publ/bcbs256.htm.

———. 2017. Basel III: Finalising Post- Crisis Reforms. Basel: Bank for International Settlements. https://www.bis.org/bcbs/publ /d424.htm.

Board of Governors of the Federal Reserve System. 2009. “Super-visory Capital Assessment Program.” Washington, DC. http://www.federalreserve.gov/bankinforeg/scap.htm.

———. 2012. “Comprehensive Capital Analysis and Review 2012: Methodology and Results for Stress Scenario Projections.” Washington, DC. http://www.federalreserve.gov/newsevents /press/bcreg/20120313a.htm.

Borio, Claudio. 2012. “The Financial Cycle and Macroeconomics: What Have We Learnt?” BIS Working Paper 395, Bank for International Settlements, Basel, Switzerland. https://www.bis .org/publ/work395.htm.

Borio, Claudio, Matthias Drehmann, and Tsatsaronis Kostas. 2012. “Stress Testing Macro Stress Testing: Does It Live Up to Expec-tations?” BIS Working Paper 369, Bank for International Settle-ments, Basel, Switzerland. https://www.bis.org/publ/work369 .htm.

Caselli, Stefano, Stafano Gatti, and Francesca Querci. 2008. “The Sensitivity of the Loss Given Default Rate to Systemic Risk: New Empirical Evidence on Bank Loans.” Journal of Financial Services Research 34 (1): 1–34.

Cerutti, Eugenio, Anna Ilyina, Yuliya Makarova, and Christian Schmieder. 2010. “Bankers without Borders? Implications of Ring- fencing for European Cross- Border Banks.” IMF Working Paper 10/247, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 / Bankers- Without- Borders- Implications- of- Ring- Fencing - for-European-Cross-Border-Banks-24335.

Čihák, Martin. 2007. “Introduction to Applied Stress Testing.” IMF Working Paper, 07/59, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2016 /12/31/ Introduction-to-Applied-Stress-Testing-20222.

Čihák, Martin, and Klaus Schaeck. 2007. “How Well Do Aggre-gate Bank Ratios Identify Banking Problems?” IMF Working Paper 07/275, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31






http://test.eba.europa.eu/web/guest/

https://www.bankofengland.co.uk/paper/2012/the-dog-and-the-frisbee.

https://www.bankofengland.co.uk/paper/2012/the-dog-and-the-frisbee.

https://www.bankofengland.co.uk/paper/2013/constraining-discretion-in-bank-regulation

https://www.bankofengland.co.uk/paper/2013/constraining-discretion-in-bank-regulation

https://www.imf.org/exter





https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/





https://www.bis.org/bcbs/qis/qis5.htm.



https://www.bis.org/publ/bcbs256.htm.��



http://www.federalreserve.gov/bankinforeg/scap.htm.

http://www.federalreserve.gov/bankinforeg/scap.htm.

http://www.federalreserve.gov/newsevents/press/bcreg/20120313a.htm

http://www.federalreserve.gov/newsevents/press/bcreg/20120313a.htm








https://www.imf.org/en/Publications/WP/Issues/2016/12/31







https://www.imf.org/exter

http://test.eba.europa.eu/web/guest/


Ong, Li Lian, editor. 2014. A Guide to IMF Stress Testing: Methods and Models. Washington, DC: International Monetary Fund.

Ong, Li Lian, and Ceyla Pazarbasioğlu. 2014. “Credibility and Crisis Stress Testing.” International Journal of Financial Studies 2 (1): 15–81. https://www.mdpi.com/2227-7072/2/1/15.

Schmieder, Christian, Heiko Hesse, Benjamin Neudorfer, Claus Puhr, and Stefan W. Schmitz. 2012. “Next Generation System- Wide Liquidity Stress Testing.” IMF Working Paper 03/12, International Monetary Fund, Washington, DC. https://www .imf.org/external/pubs/cat/longres.aspx?sk=25509.0.

Schmieder, Christian, Claus Puhr, and Maher Hasan. 2011. “Next Generation Balance Sheet Stress Testing.” IMF Working Paper 11/83, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Next - Generation-Balance-Sheet-Stress-Testing-24798.

Schmieder, Christian, and S. Philipp Schmieder. 2011. “Impact of Legislation on Credit Risk— Comparative Evidence from the United States, the United Kingdom, and Germany.” IMF Work-ing Paper 11/55, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 / The- Impact- of- Legislation- on- Credit- Risk- Comparative- Evidence- From-the-United-States-the-24703.

Stoffberg, Hesti Jacomina, and Gary van Vuuren. 2016. “Asset Correlations in Single Factor Credit Risk Models: An Empiri-cal Investigation.” Applied Economics 48 (17): 1602–17.

Taleb, Nassim N. 2010. The Black Swan— The Impact of the Highly Improbable. Revised Edition. New York: Random House.

Taleb, Nassim N., Elie Canetti, Tidiane Kinda, Elena Loukoianova, and Christian Schmieder. 2012. “A New Heuristic Measure of Fragility and Tail Risks: Application to Stress Testing.” IMF Working Paper 12/216, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2016 /12/31/ A- New- Heuristic- Measure- of- Fragility- and- Tail- Risks -Application-to-Stress-Testing-26222.

———. 2011b. Global Financial Stability Report, September 2011—Grappling with Crisis Legacies, Chapter 3. Washington, DC, Sep-tember. https://www.imf.org/en/Publications/GFSR/Issues/2016 /12/31/ Global- Financial- Stability- Report- September- 2011 -Grappling-with-Crisis-Legacies-24745.

———. 2011c. “Germany: Technical Note on Stress Testing.” IMF Country Report 11/37, Washington, DC. http://www .imf.org/en/Publications/CR/Issues/2016/12/31/Germany -Technical-Note-on-Stress-Testing-25461.

———. 2016. Global Financial Stability Report— Fostering Stabil-ity in a Low- Growth, Low- Rate Era, Chapter 1. Washington, DC, October. https://www.imf.org/en/Publications/GFSR/Issues /2016/12/31/ Fostering- Stability- in- a-Low-Growth-Low-Rate -Era.

Jobst, Andreas A., Li Lian Ong, and Christian Schmieder. 2013. “A Framework for Macroprudential Bank Solvency Stress Test-ing: Application to S- 25 and Other G- 20 Country FSAPs.” IMF Working Paper 03/12, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ A- Framework- for- Macroprudential- Bank - Solvency- Stress- Testing- Application- to- S-25-and-Other-G -40390.

Lee, Shih- Cheng, Chien- Ting Lin, and Chih- Kai Yang. 2011. “The Asymmetric Behavior and Procyclical Impact of Asset Correlations.” Journal of Banking and Finance 35 (10): 2559–68.

Lo, Andrew W. 2012. “Reading About the Financial Crisis: A 21-Book Review.” Journal of Economic Literature 50 (1): 151–78.

Lopez, Jose A. 2004. “The Empirical Relationship between Aver-age Asset Correlation, Firm Probability of Default, and Asset Size.” Journal of Financial Intermediation 13 (2): 265–283.

Marcucci, Juri, and Mario Quagliariello. 2009, “Asymmetric Ef-fects of the Business Cycle on Bank Credit Risk.” Journal of Banking and Finance 33 (9): 1624–35.

Moody’s. 2016. “Corporate Default and Recovery Rates, 1920–2015.” Special Comment, February, New York.


https://www.mdpi.com/2227-7072/2/1/15

https://www.imf.org/external/pubs/cat/longres.aspx?sk

https://www.imf.org/external/pubs/cat/longres.aspx?sk










http://www.imf.org/en/Publications/CR/Issues/2016/12/31/Germany-Technical-Note-on-Stress-Testing-25461.













CHAPTER 8

Bank Solvency and Funding Cost: New Data and New Results

STEFAN W. SCHMITZ • MICHAEL SIGMUND • LAURA VALDERRAMA

This chapter presents new evidence on the empirical relationship between bank solvency and funding costs. Building on a newly constructed dataset drawing on supervisory data for 54 large banks from six advanced countries over 2004–13, the chapter uses a simultaneous equation

approach to estimate the contemporaneous interaction between solvency and liquidity. The results show that liquidity and solvency interactions can be more material than suggested by the existing empirical literature. A 100-basis-point increase in regulatory capital ratios is associated with a decrease of bank funding costs of about 105 basis points. A 100-basis-point increase in funding costs reduces regulatory capital buffers by 32 basis points. The chapter also finds evidence of nonlinear effects between solvency and funding costs. Understanding the impact of solvency on funding costs is particularly relevant for stress testing. The analysis suggests that neglecting the dynamic features of the solvency- liquidity nexus in the 2014 EU- wide stress test could have led to a significant underestimation of the impact of stress on bank capital ratios.

position going forward, paving the way for adverse dynam-ics. The magnitude of this effect is likely to depend on the bank’s behavioral reaction to rising funding costs. On the one hand, it may react by setting higher lending rates to its borrowers. Yet this action reduces the bank’s market share and its franchise value. On the other hand, the bank might not be able to pass through additional funding costs to new lending so its internal capital generation capacity is reduced. Even if some pass- through is possible, the erosion of profits is likely to be substantial given the shorter time to repricing of liabilities relative to assets with the margin impact on the carrying values of assets outweighing that of new asset generation.2

The dynamics of adverse economic conditions on banks’ capital position can be examined through a stress testing ex-ercise. Typically, bank stress tests measure the resilience of banks to hypothetical adverse scenarios. While stressed con-ditions capture a deterioration of banks’ economic condi-tions, such as a severe recession and a sharp correction in asset prices, they do not reflect the gradual increase in fund-ing costs that banks experience as their capital buffers are

1. INTRODUCTIONThe global financial crisis appears to have been a liquidity crisis, not just a solvency crisis.1 Yet the failure to adequately model interlinkages and the nexus between solvency risk and liquidity risk led to a dramatic underestimation of risks. Liquidity risk manifests primarily through a liquidity crunch as firms’ access to funding markets is impaired, or a pricing crunch as lenders are unwilling to lend unless they receive much higher spreads. The chapter extracts funding liquidity risk from observing the costs that banks are re-quired to pay to secure market liquidity. A sudden increase in bank funding costs can have an adverse impact on finan-cial stability through the depletion of banks’ capital buffers. To preserve financial stability, it is important to assess banks’ vulnerability to changes in funding costs. The reason is two-fold. First, to the extent funding costs reflect counterparty credit risk, it is of particular interest for supervisors to deter-mine the level of capital buffers that should be held to keep funding costs at bay if and when market conditions deterio-rate. Second, funding costs are linked not only to banks’ ini-tial capital position but they also determine their capital

This chapter is based on IMF Working Paper 17/116 (Schmitz, Sigmund, and Valderrama 2017). Similar results with an extended data set and additional robustness checks can be found in Schmitz, Sigmund, and Valderrama (forthcoming).1 Shleifer and Vishny (2011) argue that liquidity problems caused by fire sales contributed to the depth and propagation of the crisis.2 This conjecture also holds, if the share of variable rate loans is high. The variable rates usually vary with market rates (for example, three-month LIBOR)

plus a fixed margin. This does not allow banks to adjust variable rates to bank- specific increases in funding costs. Similarly, interest rate hedges insure against movements in market rate, but not in bank- specific markups on market rates.


156

are likely to underestimate the negative impact of funding costs on solvency. On the other hand, OLS estimates can overstate this negative relationship if positive shocks to sol-vency, which are likely to also affect funding costs, remain unobserved. Concretely, if markets expect that a strong bank will become safer by raising its capital ratio, current funding costs might decline more than warranted by its current capi-tal position. But if this expectation is unobserved, then OLS estimates will overstate the negative relationship between solvency and funding costs. The results provide evidence that OLS underestimates the impact of capital on funding costs. Whereas a multivariate OLS- based panel regression on the dataset yields a positive relationship between banks’ capital position and funding costs, the simultaneous equation- based analysis suggests a large negative impact of capital on the cost of funding.

The results suggest more sizable effects than those found in the literature. The study finds that a 100-basis-point in-crease in regulatory capital is associated with a 105-basis- point decrease in funding costs, which is a large effect rela-tive to the existing literature, where the effect tends to be smaller, at an average of 50 basis points.7 An application of the empirical work is illustrated to inform stress testing pro-jections of bank capital ratios under stressed conditions, us-ing the 2014 EU- wide stress test exercise.

The rest of the chapter is structured as follows. Section 2 reviews the existing literature. Section 3 introduces the new dataset and presents the econometric approach. Sec-tion 4 shows the main findings on the interaction between regulatory capital and funding costs. Section 5 explores the robustness of the results to a market- based definition of bank solvency, and to banks’ bearing capacity for liquidity risk. Section 6 illustrates the dynamic impact of the solvency- funding interaction in a stress testing framework. Section 7 concludes with some policy implications.

2. RELATED LITERATUREThis chapter is related to the empirical literature on the rela-tionship between bank solvency and funding conditions, where funding conditions are defined in terms of funding costs rather than in relation to bank access to funding mar-kets.8 There are two main strands of literature: a broader set of papers seeking to explain the effect of banks’ balance sheet fundamentals on funding costs, and an emerging lit-erature examining the two- way interaction between bank solvency and the cost of funding.

7 A recent study by Aymanns and others (2016) finds that a solvency shock of 500 basis points lead to an average increase in interbank fund-ing cost of about 20 basis points, with a peak impact of 40 basis points in 2007.

8 Whereas the chapter’s baseline specification focuses on the cost of bank funding, robustness checks to include stress conditions on funding vol-umes are also included.

depleted. The analysis presented in this chapter suggests that stress test models that do not consider the dynamics between solvency and funding costs are likely to underestimate the impact of stress on bank solvency and financial stability.3 First, higher funding costs erode bank capital buffers in the short term due to the back- book effect.4 Second, capital buf-fers are further depleted in the long term as risk- sensitive in-vestors’ demand for a higher compensation to bear risk sets off adverse dynamics and lengthens the persistence of fund-ing shocks.

This chapter aims to answer two questions. First, what is the magnitude of the interaction between funding costs and solvency? Second, how can the estimated effects be used for stress testing purposes? To address these two issues, a new dataset and test for the importance of the two- way interac-tion between funding conditions and bank solvency is con-structed. The results lend support to the joint determination of funding costs and bank solvency. The chapter also pro-vides some evidence of nonlinear interactions between fund-ing costs and solvency risk, and finds that this relationship has not changed significantly during the crisis.

While these results are somewhat consistent with the lit-erature on bank solvency and funding costs, the study ex-tends the literature in two directions. First, it builds a unique dataset consisting of supervisory reporting data of 54 large banks over 2004–135 shared across supervisory agencies from six countries.6 The data is checked to ensure it is of higher quality than the publicly available sources used in other studies. Second, the study focuses on the endogenous determination of solvency and funding costs, contrary to the approach taken in most studies, which investigate funding cost drivers. To this end, the study examines the interaction between solvency and liquidity using a simultaneous equa-tion approach based on a set of exogenous instrumental vari-ables, rather than using lagged values of endogenous variables as under a vector autoregression specification. This is motivated by concern that, given the endogeneity of capi-tal and funding costs discussed previously, an ordinary least squares (OLS)-based regression is likely to yield biased coef-ficients. A priori, the direction of the bias is uncertain. On the one hand, one might argue that banks perceived by bondholders to be riskier might face both higher funding costs and hence seek to maintain higher capital ratios to ad-dress the market’s perceived risk. And if this perception is unobserved in the empirical analysis, then OLS estimates

3 With the caveat that if banks that anticipate holding riskier assets also post higher capital ratios, capital ratios would not reflect balance sheet strength but changes in riskiness of underlying assets. This would imply that we should not be able to observe a negative relationship between capital ratios and funding costs.

4 Kitamura, Muto, and Takei (2015) find that the median value of one- year- ahead pass- through for Japanese banks is 0.18.

5 Since not all of these banks are publicly traded, a restricted sample is used for some econometric specifications.

6 Due to the sensitivity of the data, strict confidentiality arrangements were in place.

Bank Solvency and Funding Cost: New Data and New Results


Stefan W. Schmitz, Michael Sigmund, and Laura Valderrama 157

solvency problems proxied by lower asset quality. The study is conducted on a panel of 7,000 US banks over the 2007–09 financial crisis.

A different estimation method is applied by Babihuga and Spaltro 2014. In the context of a panel error correction model, they estimate the long- and short- term effects of bank- specific and macro variables on funding costs using a panel of 52 banks in 14 advanced economies over 2001–12. In the long term, a 1-percentage-point increase in bank reg-ulatory capital reduces funding costs by 26 basis points, though this relationship is somewhat reversed in the short term, wherein an increase in bank capital is associated with rising bank funding costs two quarters ahead. Gray, Weh-rhahn, and Savage 2012 use a contingent claims analysis approach to compute a fair value credit default swap (FVCDS) spread as a proxy of bank funding cost using a Merton- based approach. Combining FVCDS with an im-plied market- based capital ratio, the authors find a nonlinear relationship between funding costs and bank capital. Under the baseline scenario, banks’ weighted average expected de-fault frequency (EDF) rises steadily at an accumulated pace of 75 percent by the end of the stress testing horizon. This is mapped to an equivalent 75 percent rise in FVCDS. Yet, un-der the adverse scenario, the projected accumulated increase of 150 percent in the EDF measure is linked to a larger rise in FVCDS, revealing a nonlinear relationship between market- based solvency and funding costs.

Within the second strand of the literature, Pierret 2014 uses fixed- effect panel vector autoregressive regressions to model the nexus between solvency and liquidity risk of banks in a set of 49 US banks examined over 2000 to 2013. The main result suggests an asymmetric relationship: higher solvency risk, measured by the expected capital shortfall SRISK 9 defined by Acharya and others (2010), Archarya, Engle, and Richardson (2012) and Brownlees and Engle (2011), limits the access of the firm to short- term funding. Yet a firm with more liquidity risk exposure, proxied by short- term debt, has a higher risk of insolvency in a crisis. Specifically, a unit increase in the expected capital shortfall ratio reduces its short- term debt ratio by 1.1 percentage points, suggesting that riskier banks find their access to wholesale markets limited. On the other hand, banks post-ing a 1 percent increase in short- term debt see their expected capital shortfall ratio increase by 0.9 percentage point, sug-gesting that banks funded with more short- term debt face higher solvency risk. This chapter is more closely related to Distinguin, Roulet, and Tarazi 2013, and Schmitz, Sig-mund, and Valderrama (forthcoming). Distinguin, Roulet, and Tarazi (2013) use a simultaneous equation approach to study the endogenous interaction between solvency and funding volumes on a panel of 870 US and European pub-licly traded commercial banks over 2000–06. For the

9 The SRISK measure is defined as the difference between the regulatory capital ratio applied to the expected value of assets in the event of a fi-nancial crisis and the expected market value of capital.

Within the first strand, one set of papers base their esti-mates on a multivariate panel estimation of large banks. An-naert and others 2013 find that the interaction between solvency and funding costs is indeed significant in a sample of 31 large euro-area banks over the precrisis period from 2004 through October 2008. A 1-percentage-point drop in weekly bank stock returns (associated with higher implied market- based leverage), is associated with a 64-basis-point rise in a bank’s credit default swap (CDS) spread. Similarly, Hasan, Liu, and Zhang 2016 show that solvency has signifi-cant impact on bank funding costs using a sample of 161 global banks from 23 countries over 2001–11. An increase of 1 percentage point in market- based leverage raises CDS spreads by an average of 101 basis points. This effect is slightly more pronounced after 2007 when the sensitivity of the coefficient increases to 103 basis points. In addition, they also include costs of funds (proxied by interest expense over total assets) as an explanatory variable, which turns out to be significant. However, this seems to point to an endogeneity problem, as CDS spreads and funding costs are expected to be jointly determined. Likewise, Aymanns and others 2016 examine the sensitivity of bank funding costs to bank sol-vency, drawing on the Federal Deposit Insurance Corpora-tion call report covering 10,000 banks over the period 1993–2013. They perform a panel estimation to quantify the impact of changes in bank fundamentals on yearly balance sheet measures of banks’ funding costs. The latter are cap-tured by either wholesale funding (interest rate expenses on federal funds) or average funding costs (total interest ex-pense over total liabilities). Their independent variables are bank fundamentals clustered by factor analysis. The constit-uent variables stem from four groups: solvency, liquidity, as-set quality, and profitability. They find a larger negative coefficient of bank solvency on wholesale funding costs, pointing at the higher credit risk sensitivity of wholesale in-vestors relative to depositors. Their results suggest that the sensitivity of funding cost to bank capital is larger in bad times. Whereas the average effect is typically small, with a solvency shock of 5 percentage points leading to an average increase in interbank funding cost of about 20 basis points, this effect rises to 40 basis points in 2007 when wholesale funding providers’ sensitivity to solvency risk reached its peak. The analysis also shows that the relationship between funding cost and solvency is nonlinear, with higher sensitiv-ity of funding cost at lower levels of bank solvency.

Afonso, Kovner, and Schoar 2011 conduct an event study around Lehman Brothers’ bankruptcy using transaction- level data containing all transfers by US banking institu-tions through Fedwire. They find that the worst- performing large banks access the federal funds market least, whereas small banks access the market at an increase in funding spreads of over 96 basis points. Acharya and Mora 2015 show that banks’ vulnerability to liquidity risk, defined as banks’ exposure to liquidity demand risk due to credit line drawdowns and materializing in higher deposit rates, is greater in magnitude for the class of banks with greater


Bank Solvency and Funding Cost: New Data and New Results158

Construction of a New Dataset

The variables included in the new dataset were collected spe-cifically for the purpose of estimating the simultaneous in-terdependence of bank solvency and funding costs. The data consist of an unbalanced panel of 54 large banks from six countries that cover the fourth quarter of 2004 to the fourth quarter of 2013. With 33 banks in the sample, the United States is the largest contributor to the sample. The sample also includes six Austrian, six Canadian, six Dutch, and three Nordic banks. The bank data were shared among regu-latory agencies of the respective countries under strict confi-dentiality protocols and went through careful data filtering and quality checks.12

Measuring the solvency- funding cost nexus is compli-cated due to the different frequencies of regulatory data for funding costs and solvency. The frequency of the former is usually much higher (up to daily) than for the latter (usually quarterly). The empirical analysis focuses on quarterly data. Another challenge for the analysis is posed by the choice of proxies to capture funding costs and solvency risk.

Banks can refinance their operations in different funding markets by tapping retail deposits, unsecured wholesale funding (including unsecured corporate deposits as well as funds sources from money markets and bond markets), and secured funding (including repos, securities lending, and se-curitization). The analysis proxies funding costs by the mar-ginal cost of long- term, unsecured, wholesale funding. It uses the five- year senior single name CDS spread for each bank in the sample. This is a reasonable proxy, as the sample consists of large international banks where CDS liquidity is usually higher than for the average bank. Also, CDS spreads are market- implied risk- neutral probabilities, which are ob-tained under the assumption that investors are risk- neutral and desire no risk premia, and thus are immune to shifts in risk aversion sentiment.

Alternatively, secondary market spreads could be used on active bonds to approximate the cost of wholesale funding. However, time series analysis drawing on this variable is challenging, as bond features change over time (for example, face value, maturity, covenants). In contrast, time series data for CDS spreads are ready available and do not suffer from changes in the maturity structure of a bank’s debt.

Another option is to use a measure of short- term whole-sale funding costs. The analysis used the five- year fair value CDS spreads and the reason is threefold. First, bank- specific data on short- term funding costs often reflects quoted prices rather than actual transaction prices. Second, variations in counterparty risk perception often lead to a volume reaction (that is, shortening of tenors or a reduction of lines) rather than to significantly higher rates. Third, unconventional monetary policy, including full allotment and quantitative

12 Supervisory data is based on reported regulatory balance sheets and in-clude confidential supervisory information gathered through supervi-sory processes.

solvency part, they use regulatory capital ratios as proxy. On the funding side, they focus on the inverse of the net stable funding ratio and a so- called liquidity creation indicator. They show that banks creating more liquidity have lower regulatory capital levels, and banks with lower capital ratios post higher measures of liquidity transformation. Schmitz, Sigmund, and Valderrama (forthcoming) use an extended dataset with around 300 more data points before the finan-cial crisis in 2007. With these additional data, they perform further robustness checks (cross validation and separate models for higher and lower capitalized banks). Their re-sults, however, are very much in line with our finding.

The approach taken in this chapter differs insofar as it focuses on funding costs rather than on funding volumes, and in that the relationship between solvency and funding costs on a newly constructed dataset drawing on supervisory returns is investigated. This chapter also calibrates the im-pact of incorporating the solvency- funding costs’ interaction on banks’ resilience using the 2014 European Banking Au-thority (EBA) stress testing framework.

3. THE RELATION BETWEEN SOLVENCY RISK AND FUNDING COSTSTo assess the resilience of financial institutions to adverse shocks, it is important to understand the interaction be-tween solvency and funding costs. This is particularly rele-vant in the design of stress tests where different types of shocks can affect regulatory ratios for capital and liquidity simultaneously.10

A sharp rise in bank funding costs is likely to have an adverse effect on bank capital by eroding net interest in-come. Yet the channels through which funding costs affect profits are not straightforward. A bank may react by absorb-ing higher cost of funding, thus reducing its profitability. Alternatively, the bank may try to pass on the increased cost to customers by charging high lending rates on new lending. This action might also erode profitability as liabilities reprice faster than assets and the demand for new lending is de-pressed, compressing the income base.11 The effect of bank capital on funding costs is also complex due to the highly nonlinear relation between bank asset value and solvency risk due to the short- put option embedded in bank assets. Moreover, the compensation required by investors to bear solvency risk depends on scarcity effects from compressed bond issuance under stress, on investors’ funding liquidity, and on systematic risk factors. This section uses a reduced- form approach and a broad set of controls as a useful starting point for the calibration of the impact of solvency stress on bank funding costs in supervisory stress tests.

10 Cetina (2015) discusses the channels through which shocks can impact regulatory solvency and liquidity ratios simultaneously.

11 Beau and others (2014) provide a thorough discussion of the effect of a shock to bank funding costs on bank capital and financial stability.



nonfinancial companies.15 This is a consequence of the per-ceived public- good characteristics of financial stability and the ensuing specific regulatory framework banks operate in. The study captures implicit government guarantees for bank debt by including a proxy for government credit risk re-flected in its sovereign CDS spreads, as well as by consider-ing a bank’s credit rating from S&P with the uplift based on government support. The study transforms the standard rat-ing scale into a 1 (best rating or AAA) to 24 (worst rating) numerical scale (S&P). Third, the distance to default is typi-cally higher for banks than for nonfinancial firms because banks not only have to maintain minimum regulatory capi-tal ratios but also because the required capital buffer is com-mensurate with the underlying volatility of assets. In theory that should ensure that the recovery rate of a failing bank is higher than that of nonbank financial companies. Lastly, the Merton model relies on observed values of asset volatility. Yet, as attested during the global financial crisis, the under-lying bank asset volatility is unobservable and can quickly rise if bank asset values fall, which implies that the default barrier can be reached faster than implied by the Merton ap-proach. To capture the risk of underlying assets and bank capacity to generate future profits, the study includes asset quality and net interest income as regressors. In sum, there are strong arguments to suggest that the model of bank sol-vency is more complex than that of nonfinancial companies and a broader range of variables needs to be considered. To address the robustness of the study’s results to different mea-sures of bank resilience, the estimation is rerun using a market- based measure of bank default probability over five years, namely the EDF estimated by Moody’s Credit Edge.16

A wide range of bank-specific variables are considered as potential determinants of bank solvency and funding cost. Two balance sheet variables are used, which play key roles in solvency stress tests, that is, loan loss provisions (LLPs) in percent of total assets as a measure of asset quality, and net income (NI) in percent of total assets as a proxy for banks’ return on assets and its organic recapitalization capacity. Pro-visions have a direct impact on bank solvency through their effect on RWAs. However, this proxy has known shortcom-ings. Banks have some leeway in determining loan loss provi-sions and can use it as a signaling device to the market, to accommodate regulators, to smooth earnings over time, and for tax optimization purposes. In addition, regulations and accounting rules have an impact on the level and timing of the recognition of changes in banks’ capital adequacy.17 This recognition is part of the rationale for considering, as an al-ternative to the supervisory solvency ratio in Section 4, the EDF measure, which is more market oriented. The study also

15 See BCBS 2013a. 16 Moody’s uses a Merton- based model whereby the equity of a firm is

analogous to holding a call option on the firm’s assets and the required debt payment serves as the option’s strike price. See Sun, Munves, and Hamilton 2012 for further discussion of Moody’s methodology.

17 See BCBS 2015 for further details.

easing, limited the variation and information content of short- term market rates as a proxy for banks’ marginal fund-ing costs, although the impact of unconventional monetary policy in the analysis is expected to be rather limited. The measures are available to all banks in the respective econo-mies; thus, it is not expected that it will systematically affect the variation of CDS spreads across banks. Data on individ-ual emergency liquidity assistance could reduce the bank’s CDS and affect the estimates. Though central banks try to keep emergency liquidity assistance confidential, confidence is high that no bank in the sample received it.

There are several caveats associated with the use of CDS as a measure of funding costs. First, market liquidity in CDS markets might be limited for specific banks in the sample (for example, for some of the smaller European banks). To account for this unobserved heterogeneity, bank- specific fixed effects are used. Second, CDS spreads may not be rep-resentative of bank funding costs under stress if the bank is shut out of the funding market. This chapter takes the view, however, that even under this extreme scenario, they signal effectively the marginal shadow cost of funding and thus af-fect a bank’s internal fund transfer pricing. Third, CDS spreads may reflect counterparty concerns over the issuer of credit protection. Yet, in line with the aforementioned litera-ture, it is not expected that this will systematically bias CDS spreads over the sample period. In any case, to measure funding costs effectively, the actual funding structure of each bank should be considered and the cost of alternative funding sources calibrated.13

Turning to solvency risk, the link between equity and de-fault probability has been widely established in structural models of firms’ default (Merton 1974), tested empirically (Ericsson, Jacobs, and Oviedo 2009), and used as a frame-work to calibrate Basel III regulatory capital. This motivates this study’s choice of solvency risk, that is, Core Tier 1 ratio (CT1), which reflects high- quality regulatory capital relative to risk- weighted assets (RWAs).14 Yet the relationship be-tween solvency risk and capital structure is somewhat more complex in banks relative to corporate firms. First, most bank debt is short term, which introduces liquidity risk into solvency risk. This concern is addressed by introducing bank liquidity buffers as a control variable. Second, bank regula-tion and supervision, deposit guarantee schemes, and im-plicit government guarantees (including the underpriced liquidity insurance via access to central bank emergency li-quidity assistance for illiquid and often insolvent banks) suggest that the default boundaries as well as explanatory variables for bank CDS spreads also differ from those of

13 The case studies discussed in BCBS (2013a) and BCBS (2013b) provide useful illustrations of this issue.

14 The instruments included in CT1 are well comparable across jurisdic-tions, while those included in Tier 1 and Tier 2 are comprised of instru-ments that are more country specific. Core Equity Tier 1 (CET1) would be even more comparable across jurisdictions, but was introduced only recently in Basel III. Thus, CET1 data is not available for our sample period.



tives. It is worth noting that market sentiment variables are assumed to affect directly funding costs, but not CT1 sys-tematically, though an increase in the VIX could increase the underlying volatility of bank assets, particularly if banks hold large equity portfolios, impacting their RWAs. Over time, the indirect effects are captured in the simultaneous equation approach via funding costs. Finally, we add a crisis dummy (Crisis_d) that captures significant changes in the interaction between funding costs and bank solvency as well as other time- varying control variables. Market expectations regarding bank capitalization changed abruptly with Leh-man’s bankruptcy. The dummy variable is defined as 0 from the fourth quarter of 2004 to the third quarter of 2008 and as 1 from the fourth quarter of 2008 to the fourth quarter of 2013. Despite the control variables, it is possible that the in-teraction between solvency and funding costs changed over time; for example, a stronger sensitivity of wholesale inves-tors to solvency risk post- Lehman is expected. Therefore, the study also runs its equations separately for two subsamples ( pre- and post- Lehman’s default) to check for robustness.

To control for the macroeconomic environment, the study uses country- level credit growth (loan_growth) to capture loan demand in the local credit market. High private- sector credit demand can be associated with periods of high capital ratios as banks frontload increases in CT1 to fund loan growth. One might argue that weak banks may be forced to boost their regulatory capital ratios to increase their resilience. To control for deliberate management actions, some of which were required by the supervisory agency to ease systemic risk, the study constructs a dummy variable to capture large swings in regulatory capital (ΔCT1_d). Specifically, an increase of CT1 by more than 20 percent quarter- over- quarter in nomi-nal terms serves as a proxy for deliberate management action.19 This might stem from share issuance, asset sales, or public support measures. In fact, the various public interventions in the fourth quarter of 2008 seem to be well captured by this dummy. A five- year government CDS (CDS_gov) is used as a proxy of spillovers between sovereign risk and bank funding costs. Sovereign bonds constitute the safest assets in the coun-tries in the sample and corporate bonds are priced against them. Higher sovereign CDS spreads are usually associated with higher corporate bond spreads. For the banks the inter-action can be amplified via the value of implicit and explicit government guarantees. The value of the guarantees decreases with the creditworthiness of the guarantor.

The choice of instrumental variables for identification purposes in the simultaneous equation system (8.2) is of key importance. Variables that fulfill the economic precondi-tions were selected; that is, they are directly related to one endogenous variable but interact with the second one only indirectly via the first one. They fulfill the exclusion restric-tion. In line with the literature, drivers of CDS spreads in-clude proxies of profitability and asset quality. LLPs are used

19 This yields 70 observations for the dummy variable across all banks and quarters.

controls for banks’ resilience to liquidity shocks, which is monitored regularly by the regulatory authorities. Liquidity risk is defined as a bank’s liquidity risk exposure measured by its short- term wholesale debt (liabilities with a remaining ma-turity of less than three months) over its liquidity risk- bearing capacity defined as the stock of liquid assets (cash and central bank excess reserves, sovereign debt with risk weights of 0 and 20 percent). A higher ratio implies that the bank is ex-posed to higher rollover risk. Also, wholesale funding is more credit sensitive and is likely to react more strongly to an ero-sion of bank capital buffers. At the same time, banks might profit from maturity transformation to a larger extent by funding a larger share of long- term assets with short- term wholesale funding, supporting bank profitability and easing credit risk. The sign of the liquidity risk coefficient is likely to depend on the initial capital position of banks.

The cost of funding also depends on investors’ confidence in banks’ funding instruments and in changes to macroeco-nomic conditions. The study addresses the potential regime shift around the outbreak of the global financial crises in 2008 by using the following control variables. First, a non-bank, non- country- specific variable was included that prox-ies for market sentiment in the interbank market. The London Interbank Offered Rate (LIBOR)-overnight index swap (OIS) spread is a widely used gauge for tensions in money markets. It tends to be high in times of stress and low otherwise. Second, the study controls for substantial changes in monetary policies and for the introduction of unconven-tional measures that were designed to dampen bank funding costs by using the OIS as a proxy for the monetary policy stance at the global level. While the specifics of unconven-tional measures differ between the various currency areas in the sample and the reliance of individual banks on these central bank measures differ, this information is not publicly available in a systematic manner. The model allows for bank- specific fixed effects to capture such unobservable differ-ences. Third, a market measure of volatility is included to capture global risk aversion.18 This is motivated by evidence that a common systemic risk factor can reduce the discrep-ancy between modeled and actual returns for corporate bondholders (Chen, Collin- Dufresne, and Goldstein 2009). The study proxies global risk aversion by the VIX index. This is a reasonable assumption, as the sample of banks includes internationally active banks holding international asset port-folios and raising funding from international creditors. Global risk attitude can have an impact on bank funding costs, especially for hedging products such as credit deriva-

18 The high correlation between VIX and LIBOR- OIS reported in the ap-pendix is an artifact of the enormous spikes in both around the Lehman failure. Before and after, the two did not move together. They indeed measure and capture different phenomena: the VIX captures a very broad change in volatility across all sectors of the economy (macroeco-nomic news in various parts of the world, changes in risk sentiment, geopolitical tensions); the LIBOR- OIS spread captures observed price behavior in levels in a subsegment of the economy (unsecured interbank markets). It is very bank specific and time specific.



as the instrumental variable for identification in the CT1 equation in Specification 1. LLPs are a proxy for asset qual-ity and directly affect CT1, as lower credit quality increases risk-weighted assets and, thus, the denominator of the CT1 ratio. LLPs affect FVCDS only indirectly via counterparty risk, that is, indirectly via CT1. Similarly, net income NI is used as an exogenous variable in the solvency equation. The main channel through which solvency affects NI is via fund-ing costs, which are captured in the model setup. Other de-terminants of NI like commission income (fees and turnover); staff costs, IT- costs, LLP, participations, and re-turn on own portfolio are not directly affected by solvency. In addition, country- wide loan growth is included only in the CT1 equation. In Specification 2 a dummy variable is added that captures deliberate management action to change CT1 in the CT1 equation regulatory capital (ΔCT1_d). It affects FVCDS only via CT1. The S&P rating (S&P [lag 1]), the sovereign CDS spread (CDS_gov), and the LIBOR- OIS

spread for the identification of equation FVCDS are used in Specification 1. The lagged S&P rating directly affects banks’ CDS spreads; it can have an indirect impact on CT1 eventually via higher funding costs. Similarly, sovereign CDS spreads and the LIBOR- OIS spreads directly affect bank funding costs but not banks CT1 ratios.

Table 8.1 shows data coverage for the variables used in the estimation, whereas Table 8.2 presents the summary sta-tistics. Note that most of the variables are denoted in per-centage points. This also holds for CDS spreads. The median value stood at 131 basis points across all banks over the en-tire period. The quartiles of the EDF measure are: 0.08 per-cent (first), 0.3 percent (second), and 0.94 percent (third). In addition, Table 8.3 provides a cross- correlation matrix of the dependent and independent variables used in the analysis. Interestingly, regulatory and market- based measures of bank solvency are not highly correlated with a correlation coeffi-cient below 10 percent. EDF measures are more closely

TABLE 8.1

Data Coverage

VariableAvailable Observations

(Number)Available Observations

(In Percent)

Dependent variableCT1 1632 81.7EDF 1625 81.3FVCDS 1625 81.3CET1 1159 58.0Tier 1 477 23.9FVOAS 764 38.2ptb 1184 59.3tce 1458 73.0

Bank characteristicsassets_usd 1847 92.4NPL 1365 68.3LLR 1569 78.5LLP 1839 92.0LTD 1719 86.0st_debt 1565 78.3excess_reserves 1823 91.2fx liabilities 276 13.8NIE 1839 92.0NII 1815 90.8NI 1839 92.0Fitch 1311 65.6Moodys 1395 69.8S&P 1514 75.8

Country variablesER_regime 1998 100.0CDS_gov 1515 75.8loan_growth 1997 100.0

Sources: Bloomberg, LP; IMF International Financial Statistics database; Moody’s KMV; national supervi-sory data; and Thomson Reuters.Note: Coverage of key variables for the sample of European and North American banks from 2004:Q4 to 2013:Q4. assets_usd = total assets in US dollars; CDS_gov = sovereign credit default swap spread; CET1 = Core Equity Tier 1 ratio; CT1 = Core Tier 1 ratio; EDF = expected default frequency; ER_regime = IMF’s de facto classification of exchange rate arrangements (scalar); FVCDS = fair value credit default swap spread; FVOAS = fair value option adjusted spread; fx liabilities = liabilities in foreign currency to total liabilities; NI = net income; NIE = net interest expense; NII = net interest income; LLP = loan loss provi-sions; LLR = loan loss provisions to total assets; LTD = loan-to-deposit ratio; NPR = net-profit-to-total-assets ratio; ptb = price-to-tangible-book-equity ratio; tce = tangible common equity to total assets.



TABLE 8.2

Summary Statistics of the Dependent and Independent VariablesVariable Min First

QuartileMedian Mean Third

QuartileMax NAs Standard

DeviationCT1 −13.7 7.9 9.4 10.5 11.6 111.2 366 9.8EDF 0.0 0.1 0.3 0.9 0.9 21.4 373 1.6FVCDS 0.0 0.4 1.3 2.0 2.5 17.4 381 2.2∆CT12_sign −525.3 0.0 0.0 −0.1 0.1 757.8 423 28.2∆EDF2_sign −87.4 0.0 0.0 0.1 0.0 158.3 456 6.8∆FVCDS2_sign −73.3 0.0 0.0 0.1 0.1 132.0 427 5.9CET1 −13.7 6.4 7.9 7.8 10.0 19.7 839 4.0Tier 1 4.4 8.6 10.4 14.7 12.1 114.6 1299 18.1ptb 10.2 91.4 145.4 163.7 217.9 577.9 814 93.1tce 0.0 343.2 483.5 485.9 647.8 1726.0 540 249.4assets_usd 28 78 162 359 401 2460 151 480.1LLP −0.2 0.0 0.1 0.1 0.2 2.0 183 0.2NI −4.7 0.1 0.2 0.1 0.3 2.1 160 0.3Fitch 2.0 4.0 4.0 4.9 6.0 10.0 687 1.7Moodys 1.0 4.0 5.0 5.3 7.0 15.0 603 2.3S&P 2.0 5.0 5.0 5.7 7.0 11.0 484 1.8∆CT1_d 0.0 0.0 0.0 0.0 0.0 1.0 423 0.2CDS_gov 0.0 0.2 0.4 0.4 0.5 1.9 483 0.3loan_growth −7.7 −0.1 1.3 1.0 2.2 8.2 1 1.9VIX 11.0 13.6 18.3 20.5 24.3 58.3 0 9.5LIBOR_OIS 0.1 0.1 0.2 0.3 0.4 2.1 0 0.4Crisis_d 0 0 1 0.6 1 1 0 0.5

Sources: Bloomberg, L.P.; Datastream; International Financial Statistics; Moody’s KMV; and national supervisory data.Note: Summary descriptive statistics of the sample of European and North American banks from 2004:Q4 to 2013:Q4. All variables are expressed in percent, except assets in billions of US dollars, agency ratings in a numerical scale (from 1 for AAA to 24 for D), and two dummy variables, i.e. ∆CT1_d and Crisis_d (values: 0, 1). Key variables include: CT1 (Core Tier 1 to RWAs); EDF (Moody’s 5y expected default frequency); FVCDS (Moody’s 5y fair value credit spread); ∆CT12_sign (square quarter-on-quarter growth rate of CT1, sign preserving); ∆EDF2_sign (square quarter-on-quarter growth rate of EDF, sign preserving); ∆FVCDS2_sign (square quarter-on-quarter growth rate of FVCDS, sign preserving); CET1 (Core Equity Tier 1 to RWAs); Tier 1 (Tier 1 equity to RWAs); ptb (price to tangible book equity); tce (tangible common equity to total assets); assets_usd (total assets in billion USD); LLP (loan loss provisions to total assets); NI (net income to total assets); Fitch, Moody’s, S&P (agency bank’s rating with government uplift mapped to a numerical scale from 1 (AAA) to 24 (D)); ∆CT1_d (dummy variable with 1 if quarter-on-quarter growth of CT1 is >20%; 0 otherwise); CDS_gov (5y gov-ernment CDS); loan_growth (quarter-on-quarter growth of loans to the private sector); VIX (implied volatility of S&P 500 index options); LIBOR-OIS (3m libor usd to overnight index swap); and Crisis_d (dummy variable with 1 for 2008:Q4 to 2013:Q4; 0 otherwise). CDS = credit default swap; EDF = expected default frequency; FVCDS = fair value credit de-fault swap spread; max = maximum; min = minimum; NAs = number of missing observations; RWAs = risk-weighted assets.

linked to other market- based measures including CDS spreads of government bonds and S&P’s bank ratings. While the components of the profit- and- loss account are all linked in various ways, the correlation between NI and LLP at 40 percent is not particularly significant in our sample. This might be explained by the fact that there are many other determinants of NI so that the increasing LLPs do not mechanistically reduce NI. The latter is mostly determined by interest income (slope of the yield curve, bank-specific funding costs) and commission income (fees and turnover); staff costs, IT costs, LLPs, participations, return on own portfolio, and a number of other factors also play a role.20

20 Even if variables LLP and NI were collinear, the estimated coefficients would still be consistent in the study’s estimation procedure. The stan-dard error would be inflated but that would not affect the main finding, namely that solvency and funding costs are endogenously determined and that neglecting that interaction in stress tests leads to the systematic and significant underestimation of the effects on solvency of a given shock.

Potential stationarity- related concerns are addressed by performing the so- called meta unit root tests by Choi 2001, which include unit- root tests for each variable separately and tests the p- values from these tests to produce an overall re-sult. The null hypothesis of a unit root is rejected in most tests. The distribution of banks’ solvency and funding costs is shown in Figure 8.1. CT1 ratios are presented in the top chart. Over the sample period, the first quartile is 7.89 percent, the third quartile is 11.55 percent, the mean is 10.5 percent, and the median is 9.42 percent. The figure reveals banks’ efforts to build their capital buffers in the wake of the financial cri-sis with average CT1 ratios increasing almost twofold from 7.4 percent in 2007 to 13.7 percent in 2013. The distribution has widened somewhat across time and outliers on the top of the distribution have become gradually more prominent. Panels 2 and 3 display the distribution of five- year EDF and five- year CDS market- based measures. The CDS first quar-tile is located at 45 basis points, the second quartile is lo-cated at 131 basis points, and the third quartile is located at


Stefan W. Schm

itz, Michael Sigm

und, and Laura Valderrama

163

TABLE 8.3

Cross-Correlation Matrix of the Dependent and Independent Variables∆CT1_d CDS_gov CT1 Crisis_d EDF FVCDS LIBOR_

OISLiRisk loan_

growthLLP NI OIS S&P ∆CT12_

sign∆FVCDS2_

signVIX

∆CT1_d 1.00 0.00 −0.05 0.00 0.05 0.07 0.22 0.08 −0.03 0.11 −0.10 0.02 −0.03 0.04 0.03 0.18CDS_gov 0.00 1.00 0.10 0.53 0.41 0.39 0.02 −0.09 −0.31 0.05 −0.01 −0.50 0.12 0.04 0.03 0.21CT1 −0.05 0.10 1.00 0.20 0.09 0.12 −0.03 −0.02 −0.06 −0.04 −0.02 −0.19 −0.06 0.13 −0.04 0.00Crisis_d 0.00 0.53 0.20 1.00 0.40 0.48 0.11 −0.10 −0.52 0.19 −0.17 −0.91 0.26 0.04 −0.05 0.40EDF 0.05 0.41 0.09 0.40 1.00 0.86 0.12 −0.05 −0.37 0.30 −0.21 −0.38 0.28 0.03 0.19 0.26FVCDS 0.07 0.39 0.12 0.48 0.86 1.00 0.33 −0.06 −0.45 0.35 −0.32 −0.49 0.28 0.04 0.24 0.40LIBOR_OIS 0.22 0.02 −0.03 0.11 0.12 0.33 1.00 −0.03 −0.24 0.29 −0.29 −0.17 −0.09 −0.02 0.11 0.85LiRisk 0.08 −0.09 −0.02 −0.10 −0.05 −0.06 −0.03 1.00 0.06 −0.05 −0.03 0.08 −0.12 −0.01 0.00 −0.05loan_growth −0.03 −0.31 −0.06 −0.52 −0.37 −0.45 −0.24 0.06 1.00 −0.31 0.21 0.49 −0.11 0.01 −0.08 −0.43LLP 0.11 0.05 −0.04 0.19 0.30 0.35 0.29 −0.05 −0.31 1.00 −0.42 −0.20 0.26 0.01 0.02 0.36NI −0.10 −0.01 −0.02 −0.17 −0.21 −0.32 −0.29 −0.03 0.21 −0.42 1.00 0.19 −0.04 −0.01 −0.24 −0.32OIS 0.02 −0.50 −0.19 −0.91 −0.38 −0.49 −0.17 0.08 0.49 −0.20 0.19 1.00 −0.24 −0.03 0.02 −0.40S&P −0.03 0.12 −0.06 0.26 0.28 0.28 −0.09 −0.12 −0.11 0.26 −0.04 −0.24 1.00 −0.01 0.00 −0.03∆CT12_sign 0.04 0.04 0.13 0.04 0.03 0.04 −0.02 −0.01 0.01 0.01 −0.01 −0.03 −0.01 1.00 0.02 −0.01∆FVCDS2_sign 0.03 0.03 −0.04 −0.05 0.19 0.24 0.11 0.00 −0.08 0.02 −0.24 0.02 0.00 0.02 1.00 0.08VIX 0.18 0.21 0.00 0.40 0.26 0.40 0.85 −0.05 −0.43 0.36 −0.32 −0.40 −0.03 −0.01 0.08 1.00

Sources: Bloomberg, L.P.; Datastream; International Financial Statistics; Moody’s KMV; and national supervisory data.Note: Correlation matrix of key variables for the sample of European and North American banks from 2004:Q4 to 2013:Q4. All variables are expressed in percent, except assets in billions of US dollars, S&P ratings in a numerical scale (1 for AAA, and 24 for D), and the dummy variables ∆CT1_d and Crisis_d (values: 0, 1). Key variables include: ∆CT1_d (dummy variable with 1 if quarter-on-quarter growth of CT1 is >20%; 0 otherwise); CDS_gov (five-year government CDS); CT1 (Core Tier 1 to RWAs); Crisis_d (dummy variable with 1 for 2008:Q4 to 2013:Q4; 0 other-wise); EDF (Moody’s five-year expected default frequency); FVCDS (Moody’s five-year fair value credit spread); LIBOR-OIS (3m libor usd to overnight index swap); LiRisk (ratio of cash, central bank excess reserves, and sovereign debt with risk weights of 0 and 20% to short-term wholesale liabilities with remaining maturity of less than three months); loan_growth (quarter-on-quarter growth of loans to the private sector); LLP (loan loss provisions to total assets); NI (net income to total assets); OIS (overnight index swap); S&P (agency bank rating with government uplift in a numerical scale from 1 (AAA) to 24 (D)); ∆CT1²_sign (square quarter-on-quarter growth rate of CT1, sign preserving); ∆FVCDS²_sign (square quarter- on-quarter growth rate of FVCDS, sign preserving); and VIX (implied volatility of S&P 500 index options). CDS = five-year credit default swap spread; EDF = expected default frequency; FVCDS = fair value credit de-fault swap; RWAs = risk-weighted assets.



A Simultaneous Equation Approach

To capture the contemporaneous realizations of bank solvency and bank funding costs, the study estimates the solvency and funding equations using a simultaneous equa-tion panel approach. For the purpose of stress testing, it is important to account for this endogeneity to avoid the un-derestimation of a solvency shock on financial stability.

The following model is estimated:

YΓ = XB + U (8.1)

In the analysis, Y is the vector of the two endogenous vari-ables (that is, solvency and funding costs), and X is a vector of exogenous variables including bank- specific variables (to capture governance structures or business models), country- specific variables (to control for time- varying macroeco-nomic conditions), and global variables (to capture global financial conditions and investors’ risk appetite).

Rewriting (8.1) in reduced form simplifies the problem:

Y = XBΓ−1 + UΓ−1 = XΠ + V (8.2)

249 basis points. The figure reveals that market- based mea-sures for solvency and funding costs track each other quite closely, although in periods of stress, CDS spreads react more strongly than EDF measures. Interestingly, funding costs remain elevated, even after the financial crisis subsided, despite banks’ efforts to rebuild their regulatory capital ra-tios, suggesting that market- based hurdle rates may have in-creased in the wake of the crisis. This may be partly due to investors’ risk reassessment of banks’ underlying portfolios. The distribution of market- based measures has become wider relative to that for regulatory capital measures, point-ing at higher discrimination by investors across banks’ creditworthiness.

Figure 8.2 displays the geographic evolution of the aver-ages across banks of CT1, EDF, and CDS. Whereas North American banks’ funding stress has subsided in the wake of stronger regulatory capital ratios built after the crisis, Euro-pean banks have been hit by higher funding costs despite their strong capital ratios, particularly during the sovereign debt crisis in 2012, pointing at the adverse dynamics be-tween banks and sovereigns.

2.0

4.0

6.0

10.0

14.0

8.0

12.0

0.0

16.0

2004

:Q4

05:Q

105

:Q2

05:Q

305

:Q4

06:Q

106

:Q2

06:Q

306

:Q4

07:Q

107

:Q2

07:Q

307

:Q4

08:Q

108

:Q2

08:Q

308

:Q4

09:Q

109

:Q2

09:Q

309

:Q4

10:Q

110

:Q2

10:Q

310

:Q4

11:Q

111

:Q2

11:Q

311

:Q4

12:Q

112

:Q2

12:Q

312

:Q4

13:Q

113

:Q2

13:Q

313

:Q4

0.0

3.0

0.5

1.5

2.0

2.5

1.0

2. Five-Year EDF (In percent)

1. CT1 Ratio (Percent)

2004

:Q4

05:Q

205

:Q4

06:Q

206

:Q4

07:Q

207

:Q4

08:Q

208

:Q4

09:Q

209

:Q4

10:Q

210

:Q4

11:Q

211

:Q4

12:Q

212

:Q4

13:Q

213

:Q4

0

800

100

300

500

700

400

600

200

3. Five-Year FVCDS (In basis points)

2004

:Q4

05:Q

205

:Q4

06:Q

206

:Q4

07:Q

207

:Q4

08:Q

208

:Q4

09:Q

209

:Q4

10:Q

210

:Q4

11:Q

211

:Q4

12:Q

212

:Q4

13:Q

213

:Q4

Sources: National supervisory data; and Moody’s KMV.Note: Evolution of the distribution of regulatory capital measures and market-based indicators across time. Panel 1 shows the distribution of Core Tier 1 capital ratios (CT1). Panels 2 and 3 show the distribution of five-year expected default frequency (EDF) and five-year CDS spreads (CDS). The boxplots include the mean (yellow dot), the 25th and 75th percentiles (shaded areas) and the 10th and 90th percentiles (whiskers).

Figure 8.1 Cross-Sectional Distribution of Bank Solvency and Funding Costs



account for the correlation structure of errors in each struc-tural equation. Either the 2SLS or 3SLS results are reported, depending on the results of the statistical tests.

The statistical justification of the estimation approach can be tested by a series of standard tests in the context of 2SLS and 3SLS. First, the relevance of the instruments must be tested to avoid the weak instrument problem (see Staiger and Stock 1997 for more details). For each specification, the F- statistic and the p- value are reported, testing the joint rele-vance of the instruments for each equation. Second, instrument exogeneity is tested for with two tests: the J- test is performed for each equation to check for exogeneity of the instruments Bhargava (1991). Also, the Lagrange multiplier test (LMF) suggested by Kiviet (1986) is applied. If the null hypothesis is not rejected for at least one equation in the system, these tests support the application of 2SLS as an instru -mental variable estimator. Third, endogeneity of the ( right- hand side) solvency and liquidity variables is tested for. Here, the analysis does not use the classical Hausman test that tests if all coefficients of two estimators (2SLS vs. OLS) are differ-ent. Instead the regression- based Durbin-Wu-Hausman test that tests whether the coefficients of the ( right- hand side)

Statistically, several conditions need to hold in order to extract the matrices B and Γ from the estimated matrix Π, that is, to solve the identification problem. If it is possible to deduce the structural parameters in equation (8.1) from the reduced form parameters in equation (8.2), then the model is identified. To identify the two endogenous variables, at least two exogenous sources of variation in bank solvency and funding costs need to be found. Then, two- and three- stage least squares can be applied. The two- stage least squares (2SLS) procedure has two steps. For each structural equa-tion in (8.1), each dependent variable is regressed on all ex-ogenous variables in the system and the predicted values are obtained for them.21 In the second step, the other dependent variable is regressed on the predicted value of the first depen-dent variable and on the remaining exogenous variables in the particular equation. The three- stage least squares (3SLS) combines the 2SLS with seemingly unrelated regressions to

21 It is important to note that 2SLS in a simultaneous equation system has an important advantage over the classical single equation IV instrumen-tal variable estimator: it does not use instruments that are outside of the system (that is, is not an exogenous variable in one of the equations).

Core Tier 1 in percent of RWA EDF five-year FVCDS five year

0

16

2

6

10

14

8

12

4

1. Time Average (Percent)

2004

:Q4

05:Q

205

:Q4

06:Q

206

:Q4

07:Q

207

:Q4

08:Q

208

:Q4

09:Q

209

:Q4

10:Q

210

:Q4

11:Q

211

:Q4

12:Q

212

:Q4

13:Q

213

:Q4

0

20

2

6

10

18

8

1416

12

4

3. Time Average Europe

2004

:Q4

05:Q

205

:Q4

06:Q

206

:Q4

07:Q

207

:Q4

08:Q

208

:Q4

09:Q

209

:Q4

10:Q

210

:Q4

11:Q

211

:Q4

12:Q

212

:Q4

13:Q

213

:Q4

0

25

5

15

20

10

4. Time Average Euro Area

2004

:Q4

05:Q

205

:Q4

06:Q

206

:Q4

07:Q

207

:Q4

08:Q

208

:Q4

09:Q

209

:Q4

10:Q

210

:Q4

11:Q

211

:Q4

12:Q

212

:Q4

13:Q

213

:Q4

0

14

2

6

10

8

12

4

2. Time Average North America

2004

:Q4

05:Q

205

:Q4

06:Q

206

:Q4

07:Q

207

:Q4

08:Q

208

:Q4

09:Q

209

:Q4

10:Q

210

:Q4

11:Q

211

:Q4

12:Q

212

:Q4

13:Q

213

:Q4

Sources: National supervisory data; and Moody’s KMV.Note: This panel shows the evolution of solvency ratios and funding costs for the sample of European and North American banks from 2004:Q4 to 2013:Q4. The reason behind the jump in Core Tier 1 in the bottom charts in 2008:Q1 is that the data for the Core Tier 1 ratios of the Dutch banks are reported from that time onwards and the average capital ratio of these banks is higher. EDF = expected default frequency; FVCDS = fair value credit default swap; RWA = risk-weighted assets.

Figure 8.2 Evolution of Bank Solvency and Funding Costs



endogenous variable(s) are different is applied. 22 Finally, the Hausman overidentification test is applied to test the null hy-pothesis of 3SLS versus the alternative of 2SLS (provided 2SLS is validated by the exogeneity of instruments).

These estimates are compared with those obtained with a simple OLS estimator. The OLS model yields substantial bi-ases and counterintuitive results, especially for the endoge-nous variables (see analyses in Sections 4 and 5). Ultimately, the study’s approach is a balancing act between addressing the potential weaknesses of the instruments and the biases of the OLS approach. The 2SLS and 3SLS results shown in the next section yield economically more intuitive results than do the OLS results. They also appear robust across specifications including using two different measures of sol-vency. Nevertheless, the results should be interpreted with caution given the intrinsic difficulties in finding good exog-enous instruments.

4. ESTIMATION RESULTSTable 8.4 summarizes the results for the simultaneous panel estimation for the regulatory solvency measure CT1 and bank funding costs proxied by CDS spreads (in Section 5 the robustness of the results are checked by replacing the regulatory ratio by the market- based measure of solvency EDF). Table 8.4 shows results across various specifications for solvency and funding costs. For each specification, the first column shows the results of the bank solvency equation. The second column presents the results of the funding cost equation.

To explain the solvency equation, the following variables are used: loan loss provisions, net income, aggregate credit growth, and a crisis dummy. Loan loss provisions can be in-fluenced by regulatory, tax, and profit- smoothing consider-ations. Regardless of their motivation, higher provisions reduce profits and regulatory capital.23 Country- level loan growth is included as a macro control variable. If the market is growing, banks tend to increase capital to compete for market share and to protect their franchise value. A priori, the effect of loan growth on banks’ CDS spreads is ambigu-ous. On the one hand, high loan growth could be associated with low CDS spreads if it is interpreted as sign of strong market growth, solid macroeconomic fundamentals, and sound profitability. On the other hand, it can also be associ-ated with high CDS spreads when it is interpreted as a sign of low credit standards, reckless lending, and mispricing of risk. A positive effect is expected of the crisis dummy on regulatory capital. With the Lehman collapse the market ex-pectations regarding CT1 shifted from around 6 percent to

22 See Nakamura and Nakamura (1981) for more details.23 In contrast, the effect of provisions on funding costs depends on their

motivation. Whereas higher provisioning rates designed to optimize taxation can increase intertemporal profits and push down funding costs, provisions triggered by borrowers’ lower credit quality is likely to be associated with higher funding costs.

10 percent (“10 is the new 6”). Postcrisis CT1 ratios are, on average, about 323 basis points higher than they were before the crisis.

In the funding cost equation, bank net income is also in-cluded as a key determinant. Net income is expected to be associated with lower funding costs, as banks’ capacity to generate earnings and repay outstanding debt increases. Also, a set of market- based variables is included, namely the bank’s S&P’s rating, the sovereign CDS spread, the LIBOR- OIS spread, and the VIX. These variables are, however, excluded from the solvency equation, as arguably, they do not impact directly CT1 or RWA. They do so indirectly via funding costs. A bank’s ratings directly affect the pricing of its credit derivative, but not its regulatory capital.24 Higher sovereign spreads often lead to higher bank spreads as the value of the implicit government guarantee is reduced. But they do not systematically affect banks’ regulatory capital. This is be-cause bonds of the local sovereign have a zero- risk weight and are often held on hold- to- maturity portfolios. Tensions in in-terbank markets affect bank CDS spreads by rising wholesale funding costs. Finally, higher market volatility increases in-vestors’ risk premia, pushing up funding costs.

Specification 1 is the baseline specification. It yields 782 ob-servations from 38 banks. Bank funding costs are statistically and economically significantly associated with bank solvency. A 100-basis-point increase of a bank’s CDS spread is associated with a reduction of its CT1 ratio by 32 basis points.25 This re-sult is robust across specifications. Loan loss provisions are also significant; higher loan loss provisions are negatively correlated with regulatory capital. The crisis dummy is statistically sig-nificant, has the expected sign, and an economically meaning-ful magnitude. The McElroy R² is high at 90 percent. 26

The CT1 ratio is statistically significant in the bank funding cost equation A 100-basis-point higher CT1 ratio is associated with a decrease of bank funding costs by 105 basis points. This effect is robust to alternative specifica-tions. In addition, net income has a statistically and eco-nomically significant impact on bank funding costs. Sovereign risk is also significant, pointing at the existence of a sovereign- bank nexus, while the bank rating has the expected sign and is statistically significant. Tensions in

24 Ratings are eventually considering bank solvency, but ratings change infrequently and often lag CT1 changes such that we assume that they are not simultaneously determined with solvency in each quarter.

25 The 32 basis points are an average across all banks in the sample; the impact of an increase in funding costs on an individual bank depends on the share of funding that is CDS sensitive (mostly unsecured wholesale funding), the ratio of RWAs to total assets, the term struc-ture of funding, and the pass-through of higher funding costs to new assets. The banks in the sample are large internationally active banks with significant reliance on credit- sensitive funding instruments dur-ing the sample period. At the same time, competition in credit mar-kets is high, constraining banks’ ability to pass through rising funding costs to customers. On balance, the magnitude of the coefficient is regarded as plausible, taking as a benchmark the largest bank of the sample.

26 The McElroy R² provides a goodness- of- fit measure for systems of equa-tions (McElroy 1977).



tween LIBOR- OIS and VIX (Table 8.3). The crisis dummy is statistically significant, too. The McElroy R² of 81 per-cent suggests that the explanatory value of the system is high. As an additional goodness- of- fit test results are pro-vided for the adjusted R² of 82 percent, which suggests that the equation explains most of the variation in bank fund-ing costs.

the interbank market increase bank funding costs as ex-pected.27 Global risk aversion is significant, though with a negative sign, which we attribute to the correlation be-

27 The variables that serve as instruments in the funding cost equations are significant. The study concludes that, given its assumptions, the results do not reject their usefulness as instruments.

TABLE 8.4

Bank Regulatory Capital and Funding CostsSpecification 1 Specification 2 Specification 3

CT1 FVCDS CT1 FVCDS CT1 FVCDS

Endogenous variablesCT1 −1.048***

(0.273)−1.129***(0.235)

−0.848***(0.282)

FVCDS −0.320***(0.095)

−0.324***(0.086)

−0.186***(0.0719)

∆CT12_sign 0.0761(0.0544)

∆FVCDS2_sign −0.00963(0.0249)

Exogenous variables

Bank specificLLP −1.600***

(0.346)−1.593***(0.312)

−1.844***(0.386)

NI −0.144(0.174)

−0.547**(0.224)

−0.141(0.157)

−0.565***(0.141)

−0.104(0.199)

−0.627***(0.206)

S&P (lag 1) 0.379***(0.127)

0.299***(0.075)

0.326***(0.119)

∆CT1_d 0.078(0.268)

0.0352(0.285)

Country specificCDS_gov 3.707***

(0.613)4.137***

(0.407)4.073***

(0.593)loan_growth 0.005

(0.040)0.005

(0.037)0.0360

(0.0372)

Global variablesLIBOR_OIS 0.492

(0.328)0.0171***

(0.315)0.0122***(0.00470)

VIX −0.064***(0.022)

−0.0313(0.0285)

Crisis_d 3.230***(0.180)

2.264***(0.766)

3.260***(0.165)

2.97142***(0.782)

3.221***(0.188)

1.676*(0.956)

Constant 7.466***(0.881)

8.123***(2.009)

7.470***(1.007)

9.418***(2.931)

7.001***(0.935)

7.548***(2.257)

Bank FE Yes Yes Yes Yes Yes YesAdj R2 0.984 0.825 0.984 0.825 0.984 0.824Obs 782 782 772 772 772 772

McElroy R2 0.896 0.884 0.755

Source: Authors’ calculations.Note: This table shows the results of estimating the system (1) using 2SLS. The table reports the estimated coeffi-cients, t-statistics, adjusted R2, and McElroy R2. The dependent variables are regulatory capital (CT1) and 5y fair value CDS (FVCDS). The baseline specification (Specification 1) includes a set of bank-specific variables to capture asset quality (LLP), the capacity to generate organic capital (NI), and the bank rating (S&P) lagged one period to address endogeneity. Country-specific variables include the value of sovereign support from implicit guarantees (CDS_gov) and credit growth to the private sector (loan_growth). Global variables include spreads in money markets (LIBOR-OIS), investor sentiment in equity markets (VIX), and a dummy for the global financial crisis (Crisis_d). Specification 2 includes the impact of deliberate management actions to raise regulatory capital (∆CT1_d). Specification 3 includes non-linear effects of funding costs (regulatory capital) on regulatory capital (funding costs). The results are based on quarterly data from 2004:Q4 to 2013:Q4. Adj R2 = adjusted R2; Bank FE = bank-fixed effects; CDS = credit default swap; CT1 = Core Tier 1 ratio; FVCDS = fair value credit default swap spread; LLP = loan loss provisions to total assets; NI = net income; OIS = overnight index swap; VIX = Chicago Board Options Exchange Volatility Index.



bly related to the lack of sensitivity of capital requirements to rising funding costs. The additional variables leave most other coefficients basically unaffected, except for the coeffi-cient of loan loss provisions that becomes significantly higher. Schmitz, Sigmund, and Valderrama 2019 choose a different approach to account for potential non linearity be-tween funding costs and regulatory capital. They split their sample in two parts—lower and higher capitalized banks based on the CT1 values in the second quarter of 2007. They find a stronger back- book effect for the lower capitalized banks. Thus, a higher FVCDS has a more pronounced nega-tive effect on the CT1 ratio than for the better- capitalized banks where the FVCDS coefficient in the CT1 equation is closer to 0 and insignificant.

The tests’ statistics for the econometric specifications are generally satisfactory (Table 8.5). The quality- of- instruments test rejects the null of weak instruments in all equations if the contemporary S&P’s bank rating is included in the funding cost equation. Therefore, credit ratings are instrumented by their lagged value. The J- test and the LMF test fail to reject the null of exogenous instruments. The Durbin-Wu-Hausman test is consistent with the endogeneity of the ( right- hand-side) dependent variables. The system overidentification test for the 3SLS method suggests a preference for 2SLS over 3SLS (and iterated 3SLS) for Specifications 1, 2, and 3.

To gauge the direction of the likely bias of OLS due to the endogeneity of bank solvency and funding costs, three specifications using OLS (Table 8.6) were run. For Specifi-cation 1, the study obtained statistically significant coeffi-cients of 0.17 and 0.14 for the coefficients of CT1 and CDS spreads, respectively. Similar results are obtained for Specifi-cations 2 and 3. Counterintuitively, the results suggest that higher funding costs are associated with higher CT1 ratios and that higher CT1 ratios are associated with higher fund-ing costs. This reveals that without controlling for spurious correlations and unobservable shocks, OLS estimates signifi-cantly underestimate the negative relationship between funding costs and solvency. For Specification 1, the OLS co-efficient of CDS in the solvency equation suggests a positive relationship between funding costs and bank capital with an estimated coefficient of 0.14 rather than the negative impact of −0.32 estimated under the simultaneous panel approach.

5. ROBUSTNESS CHECKSThis section offers additional support for the study’s findings that solvency and funding costs are determined simultane-ously. Several robustness checks are performed using a market- based proxy for bank solvency, and introducing a measure of liquidity risk.

Introducing a Market- Based Measure of Bank Solvency

To check the robustness of the results to the solvency mea-sure, the specifications shown in the previous section using

In Specification 2, the study examines whether taking into consideration deliberate management actions to im-prove bank solvency has any impact on the results. Capital increases directly affect CT1, but systematically covary with bank CDS spreads only through changes in CT1. It turns out that the variable capturing sharp increases of capital is not statistically significant. The results for the endogenous variables and the other exogenous variables are basically un-changed; though the coefficient of CT1 in the funding cost equation is slightly higher at −113 basis points. Specification 2 was enhanced by including banks’ funding structure, as one would expect that the risk premium component in fund-ing costs increases with the funding tenor (Hull and White 2000). Ceteris paribus, the CT1 ratios of banks with larger shares of short- term funding are likely to be less affected by an increase in five- year CDS spreads than those of banks with larger shares of long- term funding. This effect was tested for by including the share of short- term debt in total assets and the interaction term between this variable and the variable FVCDS as the explanatory variable in both equa-tions. The analysis finds that the main results for the endog-enous variables are robust with respect to the signs and the significance levels. At the same time, the coefficient of FVCDS in the solvency equation increases to −1.1 from −0.32. This effect is partly counterbalanced by the positive sign of the interaction term (0.06). An increase of the FVCDS of, say, 105 basis points decreases the CT1 ratio by 100 basis points if the bank has no short- term debt at all. If short- term debt amounts to 10 per cent of total assets (the average in the sample), the effect is reduced by 60 basis points to about 50 basis points. This has roughly the same magnitude that the corresponding parameter has in Specifi-cation 2 in Table 8.4. Regarding the other parameters in the specification, the crisis dummy remains unchanged, the LLP becomes insignificant, but NI becomes significant. Regard-ing the funding equation, the parameter of the CT1 ratio increases to −0.87 from −1.13. The coefficients of the other variables (NI, S&P, CDS_gov) remain unchanged. The vari-able VIX is now insignificant, the coefficient of the LIBOR_OIS spread decreases from 1.71 to 1.03 and that of the crisis dummy from 2.97 to 1.97.

To allow for nonlinear effects, the squared values of the endogenous variables are added in Specification 3.28 These variables are calculated as squared quarter- over- quarter first differences while maintaining the direction of the change (that is, the transformation is sign preserving). These vari-ables are treated as additional endogenous variables and their fitted values of the underlying equations are included. CT1 remains significant in the funding cost equation, while CDS spreads remain significant in the solvency equation. We do not find supporting evidence of the existence of nonlinear effects between funding costs and regulatory capital, proba-

28 However, this approach is treated as first approximation, as accounting for nonlinearities in linear models is not equivalent to constructing nonlinear models of the underlying processes.



The association between the market measure of solvency and funding costs is positive and highly significant at the 1 percent level; the coefficient 1.40 is economically signifi-cant. A money market shock— as measured by a spike in the LIBOR- OIS spread— translates into an increase in bank funding costs. Global risk aversion (VIX) reduces bank funding costs. While the coefficient is statistically signifi-cant, it has the wrong sign. We attribute this to the large spikes in LIBOR- OIS when VIX also spiked during the heights of the crisis. Changes in sovereign CDS do not di-rectly affect bank funding costs. Finally, the crisis dummy is significant; after Lehman, funding costs are generally higher. The R² is high at 77 percent. The McElroy R is very high, at just under 100 percent, which suggests that the specifica-tions including market- based measures of solvency and li-quidity are less relevant than the specifications including the regulatory solvency measure CT1.

In Specification 2, the study takes into consideration whether or not deliberate management action that aims at improving bank solvency has an impact on the results. It turns out that this variable is not statistically significant. The coefficients and standard errors of the other exogenous vari-ables, LIBOR_OIS and VIX, remain largely unaffected. However, the crisis dummy is not statistically significant anymore. In addition, the coefficients and standard errors of the endogenous variables are basically unchanged.

Again, the squared changes of the endogenous variables in the current quarter are added to test for nonlinearities in Specification 3. By contrast to the results shown in Table 8.4, funding costs have a significant nonlinear impact on the

the market- based EDF measure as a proxy of bank solvency were rerun. Table 8.7 shows the results.

In Specification 1, the analysis is based on 946 observa-tions for 38 banks in six countries from the fourth quarter of 2004 to the fourth quarter of 2013. While the test for weak instruments suggests that the instruments used in Table 8.4 are weak, one would expect that the LIBOR- OIS spread and the VIX are more likely to covary with the market solvency measure than with regulatory capital. Therefore, these two variables are included in the solvency equations shown in Table 8.7. The results show that the impact of bank funding costs on the market measure of solvency is statistically and economically significant. A 100-basis-point increase of CDS spreads is associated with an average increase in the EDF of 66 basis points. The bank- specific variable- provisioning ratio and the country- specific variable loan growth are not statis-tically and economically significant. As suggested by the test for weak instruments, the market indicators LIBOR- OIS and VIX are significant in the solvency equations, too. The VIX now has the expected sign, but the LIBOR- OIS spread influences solvency negatively through funding cost, reflect-ing high correlation across markets. The crisis dummy is sta-tistically significant, but has a negative sign. This is consistent with the results in Table 8.4, as CT1 and EDF have different signs. After controlling for higher funding costs, money market conditions, and general risk aversion, EDF is some-what lower post- Lehman, pointing at the high capitalization efforts by banks covered in the sample (as demonstrated by the positive sign of the crisis dummy in the CT1 equations in Table 8.4). The R² is high at 78 percent.

TABLE 8.5

Test Results for Bank Regulatory Capital and Funding CostsSpecification 1 Specification 2 Specification 3


Quality of instruments (H0: Instruments are weak)F statistic 1059.00 81.77 1020.27 78.46 989.79 76.44p value 0.00 0.00 0.00 0.00 0.00 0.00

Exogeneity of instruments (H0: 2SLS is valid)J-test statistic 0.55 0.08 1.02 0.13 1.25 1.06p value 0.28 0.75 0.06 0.75 0.04 0.07LMF test statistic 5.16 1.96 6.54 3.27 14.21 30.12p value 0.86 0.16 0.09 0.20 0.16 0.00

Regression-based Hausman for endogeneity of specific variables (H0: Specific variables are exogenous)

t value 6.06 7.52 6.09 5.94 6.55 5.78p value 0.00 0.00 0.00 0.00 0.00 0.00

System Overid Test (provided 2SLS is valid, H0: 2SLS is preferred to 3SLS)Hansen test statistic 31.54 53.00 59.58p value 0.00 0.00 0.00

Source: Authors’ calculations.Note: This table shows the various specification tests for the results shown in Table 8.4. The analysis tests for the quality of instruments (F-test) and the exogeneity of instruments (J-test and Lagrange multiplier test). The endogeneity of the RHS endogenous variables (t-test) is tested and the Hansen system overidentification test is applied. CT1 = Core Tier 1 ratio; FVCDS = fair value credit default swap spread; LMF = language multiplier test; 2SLS = two-stage least squares.



struments in all equations. The J-test and the LMF test fail to reject the null of exogenous instruments (except for the FVCDS equation in Specification 3, which is not important since the null is kept for the EDF equation in the same speci-fication). The Durbin-Wu-Hausman test for the solvency equation is only significant at the 7 percent level in Specifi-cation 1 and not significant in Specification 2, however in-significant for the other specifications. It suggests that

solvency equation. As funding costs increase, banks’ dis-tance to default decreases, pushing up solvency risk. The coefficients of the other endogenous variables remain statis-tically and economically significant, with very similar coef-ficients. The same holds true for the coefficients of the exogenous variables.

The test statistics are generally satisfactory (Table 8.8). The quality of instruments test rejects the null of weak in-

TABLE 8.6

Bank Regulatory Capital and Funding Costs (OLS Estimation)

Specification 1 Specification 2 Specification 3


Endogenous variablesCT1 0.173***

(0.0366)0.209***

(0.0376)0.215***

(0.0382)FVCDS 0.141***

(0.0344)0.145***

(0.0347)0.171***

(0.0372)∆CT12_sign −0.00588

(0.00628)∆FVCDS2_sign −0.0153*

(0.00791)

Exogenous variables


(0.291)−2.197***(0.295)

−2.348*** (0.304)

NI 0.148(0.148)

−0.728***(0.139)

0.148 (0.149)

−0.703*** (0.138)

0.0870(0.152)

−0.702***(0.138)

S&P (lag 1) 0.0510(0.0661)

0.123*(0.0682)

0.117*(0.0685)

∆CT1_d −0.106(0.266)

−0.0988(0.265)


(0.385)3.623***

(0.395)3.617***

(0.395)loan_growth 0.129***

(0.0295)0.131***

(0.0297)0.128***

(0.0297)

Global variablesLIBOR_OIS 1.571***

(0.143)0.0077***

(0.269)0.0078***

(0.269)VIX 0.0446***

(0.0121)0.0443***

(0.0121)Crisis_d 3.159***

(0.161)−0.782***(0.238)

3.212***(0.165)

−1.271*** (0.269)

3.170***(0.166)

−1.260***(0.270)

Constant 5.826***(0.740)

1.002(0.800)

5.705***(0.850)

0.611(0.909)

5.687***(0.849)

0.571(0.910)

Bank FE Yes No No No No NoAdj R2 0.781 0.544 0.779 0.555 0.780 0.555Obs 782 782 772 772 772 772

McElroy R2 0.999 0.990 0.760

Source: Authors’ calculations.Note: This table shows the results of estimating the system (1) using OLS. The table reports the estimated coefficients, t-statistics, adjusted R2, and McElroy R2. The dependent variables are regulatory capital (CT1) and five-year fair value CDS (FVCDS). The baseline specification (Specification 1) includes a set of bank-specific variables to capture asset quality (LLP), the capacity to generate organic capital (NI), and the bank rating (S&P) lagged one period to address endogeneity. Country-specific variables include the value of sovereign support from implicit guarantees (CDS_gov) and credit growth to the private sec-tor (loan_growth). Global variables include spreads in money markets (LIBOR-OIS), investor sentiment in equity markets (VIX), and a dummy for the global financial crisis (Crisis_d). Specification 2 includes the impact of deliberate management actions to raise regulatory capital (∆CT1_d). Specification 3 in-cludes nonlinear effects of funding costs (regulatory capital) on regulatory capital (funding costs). The results are based on quarterly data from 2004:Q4 to 2013:Q4. Adj = adjusted; Bank FE = bank-fixed effects; CDS_gov = sovereign credit default swap spread; CT1 = Core Tier 1 ratio; FVCDS = fair value credit default swap spread; LLP = loan loss provisions to total assets; NI = net income; Obs = observations; OIS = overnight indexed swap; OLS = ordinary least squares; S&P = Standard & Poor’s; VIX = Chicago Board Options Exchange Volatility Index.



TABLE 8.7

Market-Based Bank Solvency and Funding CostsSpecification 1 Specification 2 Specification 3

EDF FVCDS EDF FVCDS EDF FVCDS

Endogenous variablesEDF 1.403***

(0.119)1.346***

(0.143)1.644***

(0.110)FVCDS 0.659***

(0.0508)0.588***

(0.0492)0.549***

(0.0285)∆EDF²_sign 0.0440***

(0.0160)∆FVCDS²_sign −0.0164*

(0.00875)

Exogenous variables

Bank specificLLP 0.0754

(0.120)0.190

(0.124)0.143

(0.123)NI −0.108

(0.0679)0.123

(0.113)−0.119(0.0727)

0.0675(0.125)

−0.209**(0.0823)

0.374**(0.147)

S&P 0.0244(0.0346)

0.0865*(0.0449)

0.0559(0.0379)

∆CT1_d −0.00478(0.0371)

0.0303(0.0720)

Country specificCDS_gov 0.180

(0.266)0.634*

(0.360)0.187

(0.308)loan_growth −0.00908

(0.0126)−0.0228*(0.0125)

−0.00360(0.00926)

Global variablesLIBOR_OIS −0.0184***

(0.00126)0.0268***

(0.00217)−0.0180***(0.00142)

0.0268***(0.00269)

−0.0181***(0.00141)

0.0311***(0.00290)

VIX 0.0538***(0.00512)

−0.0760***(0.0110)

0.0557***(0.00592)

−0.0735***(0.0141)

0.0608***(0.00600)

−0.101***(0.0137)

Crisis_d −0.258***(0.0770)

0.360***(0.117)

−0.0622(0.0986)

−0.0892(0.188)

−0.0557(0.103)

0.00788(0.203)

Constant −1.039***(0.141)

1.396***(0.333)

−1.969***(0.430)

2.933***(0.616)

−1.911***(0.435)

3.225***(0.791)

Bank FE Yes Yes Yes Yes Yes YesAdj R² 0.782 0.774 0.785 0.776 0.771 0.559Obs 946 946 773 773 771 771

McElroy R² 0.999 0.990 0.723

Source: Authors’ calculations.Note: This table shows the results of estimating the system (1) using 3SLS. The table reports the estimated coefficients, t-statistics, adjusted R2, and McElroy R2. The dependent variables are market-based capital proxied by the five-year expected default frequency estimated by Moody's (EDF) and five-year fair value CDS (FVCDS). The baseline specification (Specification 1) includes a set of bank-specific variables to capture asset quality (LLP), the capacity to generate organic capi-tal (NI), and the bank rating (S&P) lagged one period to address endogeneity. Country-specific variables include the value of sovereign support from implicit guarantees (CDS_gov) and credit growth to the private sector (loan_growth). Global variables include spreads in money markets (LIBOR-OIS), investor sentiment in equity markets (VIX), and a dummy for the global financial crisis (Crisis_d). Specification 2 includes the impact of deliberate management actions to raise regu-latory capital (∆CT1_d). Specification 3 includes nonlinear effects of funding costs (market-based capital EDF) on market-based capital EDF (funding costs). The results are based on quarterly data from 2004:Q4 to 2013:Q4. Adj = adjusted; Bank FE = bank-fixed effects; EDF = expected default frequency; FVCDS = fair value credit default swap spread; LLP = loan loss provisions; OIS = overnight index swap; LMF = Lagrange multiplier test; NI = net income; Obs = observations; 2SLS = two-stage least squares; S&P = Standard & Poor’s; VIX = Chicago Board Options Exchange Volatility Index.

endogeneity is less of an issue for the market- based solvency measure. The system overidentification test is satisfactory for 3SLS across all specifications.

The direction of the bias generated is assessed by running an OLS regression on the market- based solvency measure. Re-sults are reported in Table 8.9. In line with the results ob-tained for the regulatory capital measure, OLS coefficients underestimate the impact of solvency risk on funding costs across all specifications, albeit to a smaller extent. For Specifi-

cation 1, the OLS coefficient of CDS in the solvency equation at 0.59 lies below the 0.66 estimate under the simultaneous panel approach.

Introducing a Measure of Liquidity Risk

Funding costs are likely to be determined by banks’ expo-sure to liquidity risk as recently shown by Acharya and Mora 2015. Changes in the maturity or composition of banks’



of an economically and statistically significant interaction. They find that a 100-basis-point increase in marginal funding costs is associated with a 155-basis-point increase in solvency risk; a 100-basis-point increase of the T1 ratio is associated with a 77-basis-point increase of marginal funding costs. Their findings are robust under various subsamples, alterna-tive specifications, and alternative proxies for solvency. Simi-larly, IMF 2018 presents the results of an exploratory study into the solvency- liquidity interaction in stress tests. Based on three different approaches, the study confirms that the inter-action is relevant for stress tests. It also finds that the back- book effect increases with banks’ maturity mismatch between assets and liabilities and with the share of unsecured wholesale funding; it decreases with banks’ risk density (measured as RWA over total assets) and their pass- through rates of higher bank- specific funding spreads to new loans.

6. APPLICATION TO STRESS TESTINGThis section illustrates the relevance of the empirical analysis for stress testing. The estimated relationship between sol-vency and funding costs are applied to project banks’ capital ratios under stress. The objective of stress testing is to assess banks’ resilience to adverse macroeconomic developments. While banks are routinely required to incorporate funding cost projections in their stress testing submissions, these are typically driven by risk factors linked to the scenario, notably the macroeconomic environment and the evolution of bench-mark rates, but less so to idiosyncratic risk linked to banks’ capital positions under stress. The aim of this section is to provide an estimate of the additional impact of endogenizing the solvency- funding cost channel on banks’ capital ratios in a stressful environment. The analysis is based on the adverse

funding can have important implications for measures of de-fault risk. To address this concern, a measure of liquidity risk is introduced to control for banks’ liquidity- risk- bearing capacity. For the baseline specification using the regulatory capital measure, the results are shown in Table 8.10. The im-pact of regulatory capital on funding costs is robust to the introduction of the liquidity measure. The coefficient de-creases just slightly from 1.048 to 1.028 but remains statisti-cally significant at the 1 percent confidence level (Table 8.11). Table 8.12 reports the results for the market- based solvency measure. The liquidity indicator is not statistically signifi-cant across specifications. Again, the coefficient of baseline regressors is stable, with the impact of EDF slightly decreas-ing from 1.40 to 1.37 and remaining statistically significant at the 1 percent confidence level (Table 8.13).

A caveat of the analysis is that a bank’s exposure to other risks may affect its liquidity. Any exposure may expose a bank to multiple risks and can erode a bank’s liquidity posi-tion or affect its funding costs, thereby increasing its liquid-ity risk.

Overall, our finding that the interaction between solvency and liquidity funding costs is economically and statistically significant is robust. The exact level of the parameters of sol-vency and funding costs should be investigated further for different samples of banks. Our sample consists of very large internationally active banks with a relatively short maturity structure, relatively high CDS-sensitive funding structure, and a relatively low risk density. Aldasoro and Park 2018 apply and refine our approach to a proprietary balance sheet data for 13 Korean banks covering the period from the first quarter of 2015 to the second quarter of 2015. The data includes a spe-cific reporting item, “marginal funding costs,” which the au-thors use instead of the CDS spread. They confirm our finding

TABLE 8.8

Test Results for Market-Based Bank Solvency and Funding CostsSpecification 1 Specification 2 Specification 3




Regression-based Hausman for endogeneity of specific variables (H0: Specific variables are exogenous)t value −1.83 −3.87 −0.25 −2.95 3.89 −3.44p value 0.07 0.00 0.80 0.00 0.00 0.00

System Overidentification Test (provided 2SLS is valid, H0: 2SLS is preferred to 3SLS)Hansen test statistic 4.96 5.88 0.01p value 0.17 0.21 0.99

Source: Authors’ calculations.Note: This table shows the various specification tests for the results shown in Table 8.6. We check for the quality of instruments (F-test) and the exogeneity of instruments (J-test and Lagrange multiplier test). The analysis tests for the quality of instruments (F-test) and the exogeneity of instruments (J-test and Lagrange multiplier test). The endogeneity of the RHS endogenous variables (t-test) is tested and the Hansen system overidentification test is applied. EDF = expected default frequency: FVCDS = fair value credit default swap spread; LMF = Lagrange multiplier test; SLS = two-stage least squares.



TABLE 8.9

Market-Based Bank Solvency and Funding Costs (OLS Estimation)Specification 1 Specification 2 Specification 3



(0.0279)0.971***

(0.0299)1.026***

(0.0330)FVCDS 0.591***

(0.0163)0.607***

(0.0181)0.610***

(0.0195)∆EDF2_sign −0.0173***

(0.00430)∆FVCDS2_sign −0.00185

(0.00397)

Exogenous variables

Bank specificLLP 0.351**

(0.141)0.351**

(0.155)0.333**

(0.160)NI −0.101

(0.0687)−0.114(0.0839)

−0.0782(0.0753)

−0.149*(0.0897)

−0.0856(0.0771)

−0.174*(0.0889)

S&P 0.0984**(0.0384)

0.179***(0.0428)

0.168***(0.0424)

∆CT1_d 0.215(0.143)

0.218(0.143)


(0.222)1.374***

(0.262)1.323***

(0.261)loan_growth −0.00203

(0.0141)0.00406

(0.0154)0.00371

(0.0155)


(0.115)0.0218***

(0.150)−0.0188***(0.142)

0.0216***(0.176)

−0.0188***(0.142)

0.0222***(0.176)

VIX 0.0539***(0.00517)

−0.0471***(0.00682)

0.0565***(0.00612)

−0.0419***(0.00784)

0.0565***(0.00614)

−0.0450***(0.00781)

Crisis_d −0.186***(0.0689)

0.231**(0.102)

−0.0636(0.0998)

−0.327**(0.155)

−0.0704(0.101)

−0.318**(0.153)

Constant −1.027***(0.136)

0.605**(0.256)

−2.086***(0.427)

2.374***(0.560)

−2.088***(0.428)

2.447***(0.554)

Bank FE Yes Yes Yes Yes Yes YesAdj R2 0.787 0.818 0.787 0.816 0.787 0.821Obs 946 946 773 773 771 771

McElroy R2 0.999 0.990 0.760

Source: Authors’ calculations.Note: This table shows the results of estimating the system (1) using OLS. The table reports the estimated coefficients, t-statistics, adjusted R2, and McElroy R2. The dependent variables are market-based capital proxied by the five-year expected default frequency estimated by Moody's (EDF) and five-year fair value CDS (FVCDS). The baseline specification (Specification 1) includes a set of bank-specific variables to capture asset quality (LLP), the capacity to gener-ate organic capital (NI), and the bank rating (S&P) lagged one period to address endogeneity. Country-specific variables include the value of sovereign support from implicit guarantees (CDS_gov) and credit growth to the private sector (loan_growth). Global variables include spreads in money markets (LIBOR-OIS), investor sentiment in equity markets (VIX), and a dummy for the global financial crisis (Crisis_d). Specification 2 includes the impact of deliber-ate management actions to raise regulatory capital (∆CT1_d). Specification 3 includes nonlinear effects of funding costs (market-based capital EDF) on market-based capital EDF (funding costs). The results are based on quarterly data from 2004:Q4 to 2013:Q4. Adj = adjusted; Bank FE = bank-fixed effects; CDS_gov = sovereign credit default swap spread; CT1 = Core Tier 1 ratio; FVCDS = fair value credit default swap spread; LLP = loan loss provisions to total assets; NI = net income; Obs = observations; OIS = overnight indexed swap; OLS = ordinary least squares; S&P = Standard & Poor’s; VIX = Chicago Board Options Exchange Volatility Index.

macroeconomic scenario developed by the European Central Bank for the 2014 EU- wide stress test conducted by the EBA.

To illustrate the magnitude of the interaction between solvency and funding costs on banks’ capital ratios, the study used data on European banks disseminated by the EBA on the 2014 EU- wide stress test exercise.29 The EU-

29 See EBA 2014b.

wide stress test was conducted on a sample of 124 EU banks under the assumption of a static balance sheet, which im-plies no new growth and a constant business mix and model throughout the time horizon of the exercise. The resilience of EU banks was assessed over a period of three years (2014–16).

Of the 15 EU banks covered in the sample, 11 were also included in the EU stress test exercise. The analysis focuses on this subset of banks. At the cutoff date, the aggregate



TABLE 8.10

Bank Regulatory Capital and Funding Costs (Controlling for Liquidity Risk)



Endogenous variablesCT1 −1.028***

(0.251)−1.055***(0.342)

−0.939***(0.270)

FVCDS −0.350***(0.0975)

−0.390***(0.101)

−0.219***(0.0713)

∆CT12_sign 0.0835(0.0551)

∆FVCDS2_sign −0.0125(0.0250)

Exogenous variables


(0.341)−1.432***(0.348)

−1.746***(0.385)

NI −0.194(0.173)

−0.578***(0.211)

−0.209(0.177)

−0.587***(0.215)

−0.168(0.201)

−0.635***(0.208)

S&P (lag 1) 0.281**(0.112)

0.236**(0.110)

0.293**(0.116)

LiRisk 0.0961**(0.0434)

0.0998**(0.0498)

0.0693(0.0450)

∆CT1_d 0.158(0.309)

0.0712(0.291)


(0.662)4.464***

(0.711)4.465***

(0.685)loan_growth −0.000355

(0.0407)−0.00987(0.0418)

0.0282(0.0372)

Global variablesLIBOR_OIS 0.0045

(0.303)0.0117**

(0.460)0.00766

(0.00475)VIX −0.0360

(0.0312)−0.0158(0.0269)

Crisis_d 3.254***(0.177)

2.386***(0.776)

3.287***(0.184)

2.715**(1.201)

3.236***(0.186)

1.846*(0.980)

Constant 7.550***(0.864)

8.146***(1.848)

7.693***(1.002)

8.840***(2.569)

7.119***(0.918)

8.122***(2.162)

Bank FE Yes Yes Yes Yes Yes YesAdj R2 0.987 0.845 0.987 0.846Obs 742 742 732 732 732 732

McElroy R2 0.914 0.905 0.763

Source: Authors’ calculations.Note: This table shows the results of estimating the system (1) using 2SLS. The table reports the estimated coefficients, t-statistics, adjusted R2, and McElroy R2. The dependent variables are regulatory capital (CT1) and five-year fair value CDS (FVCDS). The baseline specification (Specification 1) includes a set of bank-specific variables to capture asset quality (LLP), the capacity to generate organic capital (NI), the bank rating (S&P) lagged one period, and liquidity risk-bearing capacity (LiRisk). Country-specific variable include the value of sovereign support from implicit guarantees (CDS_gov) and credit growth to the private sector (loan_growth). Global variables include spreads in money markets (LIBOR-OIS), investor sentiment in equity markets (VIX), and a dummy for the global financial crisis (Crisis_d). Specification 2 includes the impact of deliberate management actions to raise regulatory capital (∆CT1_d). Specifi-cation 3 includes nonlinear effects of funding costs (regulatory capital) on regulatory capital (funding costs). The results are based on quarterly data from 2004:Q4 to 2013:Q4. Adj = adjusted; Bank FE = bank-fixed effects; CDS_gov = sovereign credit default swap spread; CT1 = Core Tier 1 ratio; FVCDS = fair value credit default swap spread; OIS = overnight index swap; LLP = loan loss provisions to total assets; NI = net income; Obs = observations; 2SLS = two-stage least squares; S&P = Standard & Poor’s; VIX = Chicago Board Options Exchange Volatility Index.

CET1 ratio for the study’s sample stood at 14.5, which is significantly higher than the aggregate CET1 ratio for the entire sample at 11.1 percent. At the same time, the impact of stress on bank’s capital ratios is of similar magnitude across samples: 283 basis points for the study’s subset of banks relative to 270 basis points for the entire population of banks covered in the exercise. This section addresses the

question of whether integrating second- round effects via the solvency and funding cost nexus would have had a signifi-cant impact on this capital shortfall.

The coefficients shown in Specification 1 (Table 8.4) are used to endogenize banks’ funding costs. While economet-ric results are cast in terms of CT1 rather than CET1 as the measure of regulatory capital, the undisclosed value of CT1



TABLE 8.11

Test Results for Bank Regulatory Capital and Funding Costs (Controlling for Liquidity Risk)





Regression-based Hausman for endogeneity of specific variables (H0: Specific variables are exogenous)t value 5.59 7.43 6.03 5.77 6.46 6.65p value 0.00 0.00 0.00 0.00 0.00 0.00


Source: Authors’ calculations.Note: This table shows the various specification tests for the results shown in Table 8.6. We check for the quality of instruments (F-test) and the exogeneity of instruments (J-test and Lagrange multiplier test). The analysis tests for the quality of instruments (F-test) and the exogeneity of instruments (J-test and Lagrange multiplier test). The endogeneity of the RHS endogenous variables (t-test) is tested, and the Hansen system overidentification test is applied. 2SLS = two-stage least squares; 3SLS = three-stage least squares; CT1 = Core Tier 1 ratio; FVCDS = fair value credit default swap spread; LMF = Lagrange multiplier test.

for the banks in the subsample is expected to be close to their CET1 as the weighted- sized gap between Tier 1 (a broader measure than CT1) and CET1 stood at only 100 basis points in 2013.30 The assumption is that the average funding structure of the 11 banks included in the EU stress test exercise is similar to that of the average bank in the study’s 15-bank sample.31 While this is a reasonable assump-tion given the composition of the two samples, individual results might be overestimated for banks that focus on retail funding and underestimated for banks with greater reliance on wholesale funding. The estimated relationship suggests that:

11 ,

, ,*

,

, ,

α βδ

∆ = ⋅∆ + ⋅∆∆ = ⋅∆

FVCDS CT NICT FVCDS

i t i t i t

i tfc

i t

(8.3)

where ∆CT i t1 ,* denotes bank i’s change in regulatory capital

at time t excluding the interaction effect, and ∆CT i tfc1 , denotes

the interaction effect. The marginal effect of capital (net in-come) in the funding equation is denoted by α (β), and the marginal effect of funding cost in the capital equation is de-noted by δ. Note that allowing for the interaction effect at time t carries forward to t + 1 due to its impact on ∆CT i t1 , 1

*+ and therefore on ΔFVCDSi,t+1.

30 Bank CT1 ratios are proxied by CET1 ratios, as the EBA’s CET1 pro-jections are reported under the transitional arrangements of Basel III, which are close to CT1 ratios.

31 EBA 2014b does not disclose the liability composition of the banks in-cluded in the EU- wide stress test.

11 ,

, ,*

,

, ,

α βδ

∆ = ⋅∆ + ⋅∆∆ = ⋅∆

FVCDS CT NICT FVCDS

i t i t i t

i tfc

i t

The stress test horizon is denoted by {t, t + j}. Equation (8.3) can be iterated forward to calculate the impact of the interaction effect on bank i’s capital ratio at time t + j:

1 1 1

. . . 1,

1 1,

* 1, , 1

* 1

,*

,

α δ α δ β α δ αα δ δ β{ }

{ }∆ = ⋅ ⋅∆ + ⋅ ⋅ ⋅∆ + ⋅ ⋅∆ + ⋅+ + ⋅ ⋅∆ + ⋅ ⋅∆

++ + +

+−

+ +

CT CT NI CT

CT NIi t jfc j j

i tj j

i tj j

i t

i t j i t j1 1 1

. . . 1,

1 1,

* 1, , 1

* 1, 1

,*

,

α δ α δ β α δ α β δα δ δ β{ }

{ } { }∆ = ⋅ ⋅∆ + ⋅ ⋅ ⋅∆ + ⋅ ⋅∆ + ⋅ ⋅ ⋅∆+ + ⋅ ⋅∆ + ⋅ ⋅∆

++ + +

+−

+

+ +

CT CT NI CT NI

CT NIi t jfc j j

i tj j

i tj j

i tj j

i t

i t j i t j

1 1 1

. . . 1,

1 1,

* 1, ,

,*

,

α δ α δ β αα δ δ β{ }

{ }∆ = ⋅ ⋅∆ + ⋅ ⋅ ⋅∆ + ⋅ ⋅∆+ + ⋅ ⋅∆ + ⋅ ⋅∆

++ + +

+ +

CT CT NI CT

CT NIi t jfc j j

i tj j

i tj j

i t j i t j (8.4)

Interestingly, equation (8.4) reveals a hysteresis effect of solvency shocks in banks’ capital ratios. An initial distur-bance to bank capital is long- lived due to its impact through the funding cost channel. The rate of decay is determined by the interaction between the elasticity of capital to funding costs (δ) and the elasticity of funding costs to capital (α).

Next, the analysis quantifies banks’ susceptibility to ad-verse solvency- funding cost dynamics for the selected 11 EU banks. To conduct the analysis, the individual bank projec-tions of CET1 and NI were used, as projected by the EBA in 2014, as a starting point of the iterative process. The esti-mated coefficients for NI and regulatory capital (CT1) in the funding cost equation (FVCDS) are then used to parameter-ize the adverse dynamics between bank solvency and fund-ing costs and their ultimate impact on bank’s capital ratios at the end of the stress test horizon.

In 2014, the weighted- average CET1 ratio for the study’s sample of banks decreases by 130 basis points under the ad-verse scenario, from a weighted average of 14.5 percent in 2013 to 13.2 percent in 2014. At the same time, the average NI falls by 40 basis points to –0.2 percent from 0.2 percent in 2013. Given the estimated elasticities of funding costs to



continues throughout the stress test horizon. Table 8.14 re-ports the results at the bank level for the entire stress test horizon over 2014–16.

The overall effect is significant. While macroeconomic stress reduces the aggregate capital ratio by 283 basis points over 2014–16, the overall impact, including the macro shock

CT1 and NI, the solvency shock triggers an increase in banks’ marginal wholesale funding cost of 160 basis points in 2014. This shock generates a further reduction of banks’ capital ratios by 51 basis points. The additional drop in capi-tal buffers feeds into the stress test exercise as an idiosyn-cratic funding shock the following year. This iterative process

TABLE 8.12

Market-Based Bank Solvency and Funding Costs (Controlling for Liquidity Risk)




(0.123)1.276***

(0.148)1.731***

(0.110)FVCDS 0.687***

(0.0559)0.613***

(0.0528)0.553***

(0.0272)∆EDF²_sign 0.0420**

(0.0164)∆FVCDS²_sign −0.0196**

(0.00948)

Exogenous variables

Bank specificLLP 0.0551

(0.124)0.194

(0.129)0.0542

(0.119)NI −0.0951

(0.0705)0.108

(0.115)−0.104(0.0755)

0.0290(0.127)

−0.238***(0.0871)

0.423***(0.157)

S&P 0.0166(0.0330)

0.0835*(0.0426)

0.0391(0.0345)

LiRisk −0.000471(0.00149)

−0.000592(0.00405)

−0.000666(0.0103)

∆CT1_d 0.0180(0.0434)

0.0231(0.0701)

Country specificCDS_gov 0.142

(0.302)0.749*

(0.431)−0.0972(0.363)

loan_growth −0.00660(0.0128)

−0.0221*(0.0121)

−0.00272(0.00871)


(0.00124)0.0254***

(0.00234)−0.0179***(0.00140)

0.0245***(0.00290)

−0.0181***(0.00146)

0.0318***(0.00315)

VIX 0.0507***(0.00540)

−0.0692***(0.0118)

0.0529***(0.00636)

−0.0624***(0.0152)

0.0613***(0.00638)

−0.106***(0.0148)

Crisis_d −0.264***(0.0776)

0.355***(0.122)

−0.0583(0.0992)

−0.147(0.200)

−0.0574(0.106)

0.0600(0.216)

−1.032***(0.141)

1.369***(0.335)

−2.026***(0.437)

2.894***(0.595)

−1.919***(0.446)

3.335***(0.823)

Bank FE Yes Yes Yes Yes Yes YesAdj R² 0.782 0.775 0.786 0.787 0.769 0.671Obs 905 905 733 733 733 733

McElroy R² 0.999 0.990 0.760

Source: Authors’ calculations.Note: This table shows the results of estimating the system (1) using 3SLS. The table reports the estimated coefficients, t-statistics, adjusted R2, and McElroy R2. The dependent variables are market-based capital proxied by the five-year expected default frequency estimated by Moody's (EDF) and five-year fair value CDS (FVCDS). The baseline specification (Specification 1) includes a set of bank-specific variables to capture asset quality (LLP), the capacity to gener-ate organic capital (NI), the bank rating (S&P), and liquidity risk-bearing capacity (LiRisk). Country-specific variables include the value of sovereign support from implicit guarantees (CDS_gov) and credit growth to the private sector (loan_growth). Global variables include spreads in money markets (LIBOR-OIS), investor sentiment in equity markets (VIX), and a dummy for the global financial crisis (Crisis_d). Specification 2 includes the impact of deliberate manage-ment actions to raise regulatory capital (∆CT1_d). Specification 3 includes nonlinear effects of funding costs (market-based capital EDF) on market-based capital EDF (funding costs). The results are based on quarterly data from 2004:Q4 to 2013:Q4. Adj = adjusted; Bank FE = bank-fixed effects; CDS_gov = sover-eign credit default swap spread; EDF = expected default frequency; FVCDS = fair value credit default swap spread; OIS = overnight index swap; LLP = loan loss provisions to total assets; NI = net income; Obs = observations; 2SLS = two-stage least squares; S&P = Standard & Poor’s; VIX = Chicago Board Options Exchange Volatility Index.



TABLE 8.13

Test Results for Market-Based Bank Solvency and Funding Costs (Controlling for Liquidity Risk)





Regression-based Hausman for endogeneity of specific variables (H0: Specific variables are exogenous)t value −1.89 −4.02 −0.32 −2.78 4.79 −5.24p value 0.06 0.00 0.75 0.01 0.00 0.00


Source: Author’s calculations. Note: This table shows the various specification tests for the results shown in Table 8.9. The analysis tests for the quality of instruments (F-test) and the exo-geneity of instruments (J-test and Lagrange multiplier test). The endogeneity of the RHS endogenous variables (t-test) is tested and the Hansen system overidentification test is applied. 2SLS = two-stage least squares; EDF = expected default frequency; FVCDS = fair value credit default swap spread; LMF = Lagrange multiplier; SLS = two-stage least squares.

and the adverse dynamics of the solvency- funding cost nexus slashes banks’ average capital ratio by 414 basis points. This suggests that the impact of second- round effects of the solvency- funding cost nexus might erode banks’ capital ra-tios by about half of the capital shortfall estimated by the EBA. Figure 8.3 shows the factors contributing to the shortfall in the aggregate CET1 by year. For the average bank in the sample, the interaction effect on CET1 is 51 basis point in 2014, 43 basis points in 2015, and 37 basis points in 2016. 32 As a result of the adverse reinforcing dy-namics, the relative impact of the interaction effect vis- à- vis the macro effect increases throughout the stress test horizon, from 40 percent in 2014 to over 50 percent in 2016.

The impact on capital loss in monetary units is even larger, as weaker banks tend to post higher RWAs. Overall, the interaction effect triggers a reduction in aggregate capital by €3.8 billion in 2014, €3.3 billion in 2014, and €2.8 bil-lion in 2014 for the study’s sample of banks. This represents around half of the aggregate capital losses estimated by the EBA for this subset of banks from adverse economic devel-opments over the three- year horizon.

The effect of the interaction between solvency and fund-ing costs is significant in part because the impact of funding cost rises nonlinearly over the stress test horizon. This is be-cause net income and capital deteriorate further owing to adverse reinforcing dynamics. The EBA’s stressed capital ra-

32 This estimate reflects the impact of an idiosyncratic funding shock whereby a bank’s cost of funds depends on its own capital position. By taking bank solvency into account, this element captures a key amplifi-cation channel evident during the global financial crisis. On the other hand, aggregate shocks to funding costs remain the same as under the EBA stress test scenario.

tios do include non bank- specific funding cost effects from adverse macroeconomic developments, risk aversion, and li-quidity strains, but not the bank- specific feedback effect modeled in this chapter.33 The application to stress tests shows that banks with shorter funding tenors, and/or greater reliance on CDS- sensitive funding instruments (that is, un-secured wholesale funding) and/or lower- RWA- tototal- asset ratios are more affected by the feedback effect of solvency on funding costs. The study concludes that a bank’s funding structure is not only relevant for its funding liquid-ity risk exposure, but also for its exposure to solvency shocks. On the other hand, the cumulated impact could be more severe in a tail event, as bank funding structures might be further impaired under stress. In a crisis, wholesale funding tends to shift to shorter- dated tenors, increasing the amount of liabilities that need to be rolled over at higher funding rates.

7. SUMMARY AND CONCLUSIONSWhile the existence of a relationship between bank solvency and funding costs is widely accepted in the literature, its esti-mated magnitude has been typically small. This study’s results suggest a larger impact of solvency on funding costs than sug-gested by earlier studies. The stability of the coefficients is confirmed when alternative measures of solvency risk and banks’ capacity to bear liquidity risk are considered. These new results could be in part due to the study’s newly con-structed dataset, which exploits high-quality supervisory data. They could also be driven by the econometric strategy to

33 See EBA 2014a.


Bank Solvency and Funding Cost: New

Data and N

ew Results

178

TABLE 8.14

Impact of Bank Solvency and Funding Cost Interaction—2014 EBA Stress Test(Basis Points)

Estimated Elasticities 2014 2015 2016

Bank Name Funding Costs to CET1 Funding Costs to NII CET1 to Funding Costs ∆Funding Costs ∆CET1 ∆Funding Costs ∆CET1 ∆Funding Costs ∆CET1Bank 1 −1.048 −0.547 −0.320 213 −68 296 295 308 −99Bank 2 −1.048 −0.547 −0.320 103 −33 87 228 116 −37Bank 3 −1.048 −0.547 −0.320 311 −99 310 299 557 −178Bank 4 −1.048 −0.547 −0.320 88 −28 96 231 98 −31Bank 5 −1.048 −0.547 −0.320 225 −72 91 −29 3 −1Bank 6 −1.048 −0.547 −0.320 178 −57 159 −51 107 −34Bank 7 −1.048 −0.547 −0.320 335 −107 167 −53 153 −49Bank 8 −1.048 −0.547 −0.320 59 −19 86 −28 69 −22Bank 9 −1.048 −0.547 −0.320 595 −191 858 −275 976 −312Bank 10 −1.048 −0.547 −0.320 243 −78 351 −112 506 −162Bank 11 −1.048 −0.547 −0.320 93 −30 10 −3 −42 13

Source: Authors’ calculations using European Banking Authority (EBA) stress test results and estimation results.Note: This table shows the additional impact of the interaction between funding costs and solvency ratios on banks’ Common Equity Tier I (CET1) under EBA’s adverse scenario over 2014–16. The sample of banks covers those banks included in the 2014 EU-wide exercise, which are also included in the sample. The analysis assumes constant asset size over the stress test horizon under EBA’s static balance sheet assumption.



201420152016

Shock to CET1 ratio(Basis points)

Solvency-funding nexus Macro effect Overall effect

–51

–131

–43

–85

–128

–37

–67

–104

–182

Source: Author’s calculations using European Banking Authority stress test results and estimation results.Note: CET1 = Core Equity Tier 1; EBA = European Banking Authority.

Figure 8.3 Decomposition of Impact on CET1—2014 EBA Stress Test

implement a 3SLS simultaneous equation approach, by con-trast to the OLS- based estimates that are prevalent in the ex-isting literature. Indeed, the study’s results show that OLS underestimates the solvency- liquidity interaction nexus. This might be due to investors’ expectations that a weaker bank might raise capital to rebuild its capital buffer in order to ease funding pressures or meet regulatory expectations. While a simultaneous equation approach has its own challenges re-lated to the difficulty of finding suitable instruments and avoiding overidentification, this study’s statistical tests and robustness checks provide some comfort on the estimated co-efficients in the interaction between solvency and funding costs. Still, the results should be interpreted with caution, bearing the limitations of the approach in mind.

The study’s results show that the interaction between sol-vency and funding shocks in supervisory stress test models is quantitatively relevant. The analysis suggests that, by in-corporating the dynamic interaction between solvency and funding costs in the 2014 EU- wide stress test, stressed capi-tal ratios could be depleted by a further half of the capital shortfall estimated in the original EBA analysis. This is a conservative estimate, as the EBA’s methodological approach partially incorporates rising funding costs linked to the sce-nario.34 The results are also highly relevant for cost- impact assessments of capital regulation, as the costs of higher capi-tal requirements are partly offset by lower debt servicing costs. These results provide a foundation for calibrating that effect in quantitative cost- benefit analyses of bank regula-tion. The analysis also points at the merits of incorporating solvency and liquidity interactions in the design of pruden-

34 To the extent that stressed credit spreads under the adverse scenario re-flect a weakened capital position of the banking system, the rise in wholesale funding costs projected under the EBA incorporates a “sys-temic” funding shock whereby banks’ cost of funds depends on the po-sition of the system as a whole.

tial regulation. While the study’s results are encouraging, future research should assess their robustness using larger high- quality samples and, if feasible, a broader set of instru-ments to address remaining endogeneity concerns.

REFERENCESAcharya, Viral V., Robert Engle, and Matthew Richardson. 2012.

“Capital Shortfall: A New Approach to Ranking and Regulating Systemic Risks.” American Economic Review 102 (3): 59–64.

Acharya, Viral V., and Nada Mora. 2015. “A Crisis of Banks as Li-quidity Providers.” Journal of Finance 70 (1): 1–43.

Acharya, Viral V., Lasse H. Pedersen, Thomas Philippon, and Matthew Richardson. 2010. “Measuring Systemic Risk.” Tech-nical Report, Stern School of Business, New York University, New York. Social Sciences Research Network, https://papers .ssrn.com/sol3/papers.cfm?abstract_id=1573171.

Afonso, Gara, Anna Kovner, and Antoinette Schoar. 2011. “Stressed, Not Frozen: The Federal Funds Market in the Finan-cial Crisis.” Journal of Finance 66 (4): 1109–39.

Aldasoro, Inaki, and Kyounghoon Park. 2018. “Bank Solvency Risk and Funding Cost Interactions in a Small Open Economy: Evidence from Korea.” Working Paper 738, Bank for Interna-tional Settlements. https://www.bis.org/publ/work738.pdf.

Annaert, Jan, Marc De Ceuster, Patrick Van Roy, and Cristina Ves-pro. 2013. “What Determines Euro Area Bank CDS Spread?” Journal of International Money and Finance 32: 444–61.

Arellano, Manuel, and Stephen Bond. 1991. “Some Tests of Speci-fication for Panel Data: Monte Carlo Evidence and an Applica-tion to Employment Equations.” Review of Economic Studies 58 (2): 277–97.

Arellano, Manuel, and Olympia Bover. 1995. “Another Look at the Instrumental Variable Estimation of Error- Components Mod-els.” Journal of Econometrics 68 (1): 29–51.

Aymanns, Christoph, Carlos Caceres, Christina Daniel, and Lili-ana Schumacher. 2016. “Bank Solvency and Funding Costs.” IMF Working Paper 16/64, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ Bank-Solvency-and-Funding-Cost-43792.




https://www.bis.org/publ/work738.pdf




Gorton, Gary. 2010. Slapped in the Face by the Invisible Hand. Ox-ford: Oxford University Press.

Gray, Dale, Rodolfo Wehrhahn, and Lawrie Savage. 2012. “Israel: Stress Test of the Banking, Insurance, and Pension Sectors— Technical Note.” IMF Country Report 12/88, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/Israel-Technical-Note-on-Stress-Test-of-the -Banking-Insurance-and-Pension-Sectors-25850.

Hasan, Iftekhar, Liuling Liu, and Gaiyan Zhang. 2016. “The De-terminants of Global Bank Credit- Default- Swap Spreads.” Journal of Financial Services Research August 3.

Holtz- Eakin, Douglas, Whitney Newey, and Harvey S. Rosen. 1988. “Estimating Vector Autoregressions with Panel Data.” Econometrica 56 (6): 1371–95.

Hull, John, and Alan White. 2000. “Valuing Credit Default Swaps I: No Counterparty Default Risk.” Journal of Derivatives 8 (1): 29–40.

International Monetary Fund. 2018. “Euro Area Policies: Financial Assessment Program. Technical Note— Stress Testing the Bank-ing Sector.” IMF Country Report 18/228, Washington, DC. https://www.imf.org/external/pubs/cat/longres.aspx?sk=46102.0

Kitamura, Tomiyuki, Ichiro Muto, and Ikuo Takei. 2015. “How Do Japanese Banks Set Loan Interest Rates? Estimating Pass- Through Using Bank- Level Data.” Bank of Japan Working Paper 15– E– 6, Bank of Japan, Tokyo. https://www.boj.or.jp/en /research/wps_rev/wps_2015/wp15e06.htm/.

Kiviet, Jan F. 1986. “On the Rigour of Some Misspecification Tests for Modelling Dynamic Relationships.” Review of Eco-nomic Studies 53: 241–61.

McElroy, Marjorie B. 1977. “Goodness of Fit for Seemingly Unre-lated Regressions.” Journal of Econometrics 6: 381–7.

Merton, Robert C. 1974. “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates.” Journal of Finance 29 (2): 449–70.

Moody’s Analytics. 2012. Public Firm Expected Default Frequency (EDFTM) Credit Measures: Methodology, Performance, and Model Extensions. New York: Moody’s Analytics.

Nakamura, Alice, and Masao Nakamura. 1981. “On the Relation-ships among Several Specification Error Tests Presented by Durbin, Wu, and Hausman.” Econometrica 49 (6): 1583–8.

Nickell, Stephen. 1981. “Biases in Dynamic Models with Fixed Ef-fects.” Econometrica 49 (6): 1417–26.

Pierret, Diane. 2014. “Systemic Risk and the Solvency- Liquidity Nexus of Banks.” International Journal of Central Banking 11 (3): 193–227.

Puhr, Claus, and Stefan W. Schmitz. 2014. “A View from the Pop— The Interaction between Solvency and Liquidity Stress.” Journal of Risk Management in Financial Institutions 7 (1): 38–51.

Roodman, David. 2009a. “How to Do Xtabond2: An Introduc-tion to Difference and System GMM in Stata.” Stata Journal 9 (1): 86–136.

———. 2009b, “A Note on the Theme of Too Many Instruments.” Oxford Bulletin of Economics and Statistics. 71 (1): 135–58.

Schmitz, Stefan W. 2013. “The Impact of the Liquidity Coverage Ratio (LCR) on the Implementation of Monetary Policy.” Eco-nomic Notes 42 (2): 135–70.

Schmitz, Stefan, Michael Sigmund, and Laura Valderrama. 2017. “Bank Solvency and Funding Cost: New Data and New Results.” IMF Working Paper 17/116, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues/2017 /05/15/ Bank- Solvency- and- Funding- Cost- New-Data-and-New -Results-44914.

Babihuga, Rita, and Marco Spaltro. 2014. “Bank Funding Costs for International Banks.” IMF Working Paper 14/71, International Monetary Fund, Washington, DC. https://www.imf.org/en /Publications/WP/Issues/2016/12/31/Bank-Funding-Costs-for -International-Banks-41514.

Basel Committee on Banking Supervision (BCBS). 2013a. “Li-quidity Stress Testing: A Survey of Theory, Empirics and Cur-rent Industry and Supervisory Practices.” BCBS Working Paper 24, Bank for International Settlements, Basel, Switzer-land. https://www.bis.org/publ/bcbs_wp24.htm.

———. 2013b. “Literature Review of Factors Relating to Liquid-ity Stress— Extended Version.” BCBS Working Paper 25, Bank for International Settlements, Basel, Switzerland. https://www .bis.org/publ/bcbs_wp25.htm.

———. 2015. “The Interplay of Accounting and Regulation and Its Impact on Bank Behavior: Literature Review.” BCBS Work-ing Paper 28, Bank for International Settlements, Basel, Swit-zerland. https://www.bis.org/bcbs/publ/wp28.htm.

Beau, Emily, John Hill, Tanveer Hussain, and Dan Nixon. 2014. “Bank Funding Costs: What Are They, What Determines Them, and Why Do They Matter?” Bank of England Quarterly Bulletin 2014 Q4, Bank of England, London. https://www .bankofengland.co.uk/ quarterly- bulletin/2014/q4/ bank- funding - costs- what- are- they- what- determines- them-and-why-do-they -matter.

Bhargava, Alok. 1991. “Identification and Panel Data Models with Endogenous Regressors.” Review of Economic Studies 58 (1): 129–40.

Binder, Michael, Cheng Hsiao, and M. Hashem Pesaran. 2005. “Estimation and Inference in Short Panel Vector Autoregres-sion with Unit Roots and Cointegration.” Econometric Theory 21 (4): 795–837.

Brownlees, Christian T., and Robert Engle. 2011. “Volatility, Cor-relation and Tails for Systemic Risk Measurement.” Technical Report, Stern School of Business, New York University, New York. http://scrc.scrc.nyu.edu/scrc/?p=2241.

Cetina, Jill. 2015. “Incorporating Liquidity Shocks and Feedbacks in Bank Stress Tests.” OFR Brief Series, Office of Financial Research, Washington, DC, July. https://www.financialresearch.gov/briefs /2015/07/22/ incorporating- liquidity- shocks- and-feedbacks -in-bank-stess-tests/.

Chen, Long, Pierre Collin- Dufresne, and Robert S. Goldstein. 2009. “On the Relation between the Credit Spread Puzzle and the Eq-uity Premium Puzzle.” Review of Financial Studies 22 (9): 3367–409.

Choi, In. 2001. “Unit Root Tests for Panel Data.” Journal of Inter-national Money and Finance 20 (2): 249–72.

Distinguin, Isabelle, Caroline Roulet, and Amine Tarazi. 2013. “Bank Regulatory Capital and Liquidity: Evidence from US and European Publicly Traded Banks.” Journal of Banking and Finance 37 (9): 3295–3317.

Ericsson, Jan, Kris Jacobs, and Rodolfo Oviedo. 2009. “The Deter-minants of Credit Default Swap Premia.” Journal of Financial and Quantitative Analysis 44 (2): 109–32.

European Banking Authority (EBA). 2014a. Methodological Note EU- Wide Stress Test 2014. London: European Banking Authority. https://www.eba.europa.eu/-/eba-publishes-common-methodology -and-scenario-for-2014-eu-banks-stress-test.

———. 2014b. 2014 EU- Wide Stress Test Results. London: European Banking Authority. http://www.eba.europa.eu/ risk- analysis- and -data/eu-wide-stress-testing/2014/results.


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Israel-Technical-Note-on-Stress-Test-of-the-Banking-Insurance-and-Pension-Sectors-25850




https://www.boj.or.jp/en/research/wps_rev/wps_2015/wp15e06.htm/

https://www.boj.or.jp/en/research/wps_rev/wps_2015/wp15e06.htm/



https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Bank-Funding-Costs-for-International-Banks-41514



https://www.bis.org/publ/bcbs_wp24.htm.



https://www.bis.org/bcbs/publ/wp28.htm



http://scrc.scrc.nyu.edu/scrc/?p=2241

https://www.financialresearch.gov/briefs/2015/07/22/









Sun, Zhao, David Munves, and David T. Hamilton. 2012. “Public Firm Expected Default Frequency (EDF) Credit Measures: Methodology, Performance, and Model Extensions.” Technical Report, Moody’s Analytics.

Tarashev, Nikola A. 2008. “An Empirical Evaluation of Structural Credit Risk Models.” International Journal of Central Banking 4 (1): 1–53.

Valderrama, Laura. 2017. “An Agent- Based Model for Stress Testing.” Unpublished. International Monetary Fund, Washington, DC.

Verbeek, Marno. 2012. A Guide to Modern Econometrics. Hobo-ken: John Wiley & Sons, Ltd.

Schmitz, Stefan, Michael Sigmund, and Laura Valderrama. Forth-coming. “The Interaction between Bank Solvency and Funding Costs: A Crucial Effect in Stress Tests.” Economic Notes.

Shleifer, Andrei, and Robert W. Vishny. 2011. “Fire Sales in Fi-nance and Economics.” Journal of Economic Perspectives 25 (1): 29–48.

Sims, Christopher A. 1980. “Macroeconomic Reality” Economet-rica 48 (1): 1–48.

Staiger, Douglas, and James H. Stock. 1997. “Instrumental Vari-ables Regression with Weak Instruments.” Econometrica 65 (3): 557–86.


CHAPTER 9

Sovereign Risk in Macroprudential Solvency Stress Testing

ANDREAS A. JOBST • HIROKO OURA

This chapter explains the treatment of sovereign risk in macroprudential solvency stress testing, based on the experiences in the Financial Sector Assessment Program (FSAP). Four essential steps are discussed in assessing the system- wide impact of sovereign risk: scope, loss estimation,

shock calibration, and capital impact calculation. Most importantly, a market- consistent valuation approach lies at the heart of assessing the resil-ience of the financial sector in a tail- risk scenario with sovereign distress. A flexible, closed- form approach to calibrating haircuts is presented based on changes in expected sovereign defaults affecting bank solvency during adverse macroeconomic conditions. This chapter demonstrates the ef-fectiveness of using extreme value theory (EVT) in this context, with empirical examples from past FSAPs.

This chapter is based on IMF Working Paper 19/266 (Jobst and Oura 2019). The authors are grateful to Martin Čihák, Udaibir Das, Ehsan Ebrahimy, Caio Ferreira, Jad Khallouf, Raphael Lam, James Morsink, Erlend Nier, Luc Riedweg, and Mustafa Saiyid as well as staff from the national authorities of Germany, Italy, Mexico, Spain, and Saudi Arabia for their helpful comments and suggestions. The views expressed in this chapter do not represent those of the authors’ current employers. The authors thank Carlos Caceres (IMF) for his contribution to Appendix IV as well as Pavel Lukyantsau and Xiaodan Ding for their excellent data assistance.1 Various factors could encourage banks to hold sovereign exposures that strengthen sovereign- bank linkages, including regulatory incentives (for example,

low risk weights for sovereign securities), risk- taking incentives (for example, banks invest in a higher- risk sovereign to earn attractive spreads over their funding costs), and economic cyclical factors (for example, countercyclical fiscal policies). See Appendix 9.5 for a comprehensive overview of these factors.

quent (but the effects are diversified among many counter-parties), sovereign default occurs rarely but has wide- ranging consequences. Sovereign risk also results in many hard- to- assess spillover effects across sectors and countries, which can further amplify the bank- sovereign nexus. Sovereign distress can take many forms, including (1) outright default or restructuring; (2) a technical default (for example, miss-ing payments if there is no fundamental debt sustainability problem); (3) currency redenomination; (4) hyperinflation (and currency crisis); and (5) default by quasi- sovereign enti-ties (BCBS 2017a; Ams and others 2018). These severe forms of distress have been more frequently observed in EMDEs and affect banks through direct (for example, losses from di-rect exposures) as well as indirect (for example, the impact on economic growth, inflation, and exchange rates) transmis-sion channels. The episode of such explicit sovereign distress is rare among AEs in the post- World War II period. During the European sovereign debt crisis, for instance, the key channels included valuation losses of sovereign securities

1. INTRODUCTIONSovereign distress has been a significant source of systemic financial risk in many countries where banks hold large ex-posures to the public sector. In advanced economies (AEs), bank claims on domestic government debt range from a few percent of bank assets (for example, Sweden and Switzer-land) to more than 10 percent of assets (for example, Italy, Japan, and Spain; see Figure 9.1). In many emerging market and developing economies (EMDEs), sovereign exposures are twice as high as in AEs on average and particularly large in Argentina, Brazil, China, Egypt, Hungary, India, and Mexico. A broader definition of sovereign exposures (see Box 9.1), which includes subnational governments, lending (that is, loans and receivables), and sovereign guarantees, more than doubles these amounts.1

In addition to the sheer size of exposures, sovereign risk could affect banks’ solvency through a wide range of trans-mission channels with potentially complex feedback effects. Unlike private- sector debt, where individual default is fre-


184

(BCBS 2018a, 2018b) confirm that exposures other than securities (that is, banking book exposures) are a large part of overall sovereign exposures.

The method to estimate potential losses from a sovereign distress scenario in FSAPs has been broadly following the Basel III framework, with some modifications desirable for a macroprudential perspective. The outputs of FSAP stress tests are not expected to result in immediate supervisory actions— unlike those of microprudential stress tests— but may inform discussions on the robustness of systemic crisis preparedness and the use of possible macroprudential policy measures. As a result, they focus on assessing the potential capital impact of systemic risk as fully and transparently as possible, mainly by applying market- implied estimates of ex-pected sovereign default to all types of sovereign exposures. In contrast, microprudential rules smooth out short- term cycli-cal volatility to avoid introducing excessive procyclicality to capital ratios.

The main FSAP approach for stress testing sovereign risk has been to measure valuation effects on traded government debt caused by changes in expected default rather than actual default during adverse macroeconomic conditions.2 A sover-eign risk shock is calibrated as the market- consistent haircut implied by the estimated (and not realized) decline in the fair value of government bonds (“market valuation approach”) using their price or yield volatility (for example, standard deviation). For each country, the haircut reflects the observ-able cost of protecting the value of government bonds against

2 See Jobst, Ong, and Schmieder 2013 and 2017 for a broader review of macroprudential solvency and liquidity stress testing in major FSAPs.

and their impact on bank funding costs as well as the feed-back effects to sovereigns through potential bank support measures (Enria and others 2016).

The chapter shows how to assess banks’ vulnerability to sovereign risk in macroprudential stress testing. Four aspects of the tests are discussed: scope of exposures and transmis-sion channels, loss estimation methods, shock calibration, and calculation of capital impact. The discussion is largely based on the experiences with stress testing of banks in the IMF’s Financial Sector Assessment Program (FSAP) over the past decade. The same loss- estimation- and- calibration ap-proach is, in principle, applicable to not only banks but also other types of financial institutions, such as insurance com-panies, pension funds, and asset managers.

The potential scope for sovereign risk varies across coun-tries. So far, FSAPs have focused on the direct impact through security holdings, but a more comprehensive cover-age of exposures and channels seems appropriate in some cases (IMF 2015). This is partly because the European sovereign debt crisis, which motivated the integration of sovereign risk in stress tests, took place in countries where most sovereign exposures were securities (and, thus, banks’ solvency situations were significantly influenced by the market valuation of their security holdings). However, such an approach may miss essential transmission channels in other countries where the primary bank- sovereign direct linkages stem from loans to the general government and its deposits. If a country has a relatively frequent history of sov-ereign default— including payment delays, restructuring, hyperinflation— then the test may need to incorporate these channels explicitly. Indeed, the Basel III monitoring exercises


2016 or latest available 2001 2007

1. Advanced Economies 2. Emerging Market and Developing Economies

0

50

51015202530354045

Switz

erla

nd

Italy

Irela

ndUn

ited

King

dom

Swed

enAu

stra

liaKo

rea

Neth

erla

nds

Fran

ceCa

nada

Hong

Kon

g SA

RGr

eece

Germ

any

Belg

ium

Portu

gal

Unite

d St

ates

Spai

nJa

pan

0

50

51015202530354045

Chile

Mal

aysi

aCh

ina

Russ

iaIn

done

sia

Colo

mbi

aSo

uth

Afric

aTu

rkey

Phili

ppin

esM

exic

oHu

ngar

yBr

azil

Indi

aAr

gent

ina

Egyp

t

Sources: Authors’ calculations, Haver Analytics, IMF International Financial Statistics.Note: The charts show sovereign exposures as bank claims on central, state, and local governments, except for China, Hong Kong SAR, India, Switzerland, and the United Kingdom, where data on claims on state and local governments are not available. Data for the United Kingdom are missing for the years 2001 and 2007. For Egypt and Turkey, missing data for 2001 were replaced with data for 2002 and 2004, respectively.

Figure 9.1 Bank Sovereign Exposures (Bank claims on domestic government debt in percent of total banking sector assets)


Andreas A. Jobst and Hiroko Oura 185

rising default risk perceived in markets and is applied to most sovereign security holdings to determine the capital impact of the sovereign risk shock.

In this chapter, the existing approach toward a tractable method for the calibration of sovereign risk shocks as tail events is advanced. For instance, if a shock is defined relative to the historical average, the shape of the distribution func-tion is fundamental to the calibration process. For instance, a two- standard- deviation shock drawn from a standard normal distribution is much smaller than if the shock were drawn from a distribution with a fat tail. The size of the haircut could differ substantially depending on the method to account for tail risks. For instance, a one-standard- deviation- valuation shock from a distribution with a fat tail is far more sizeable than the shock drawn from a standard normal distribution. Therefore, a generalized extreme value (GEV) distribution is fitted to the historical spread dynamics of spot and forward sovereign credit default swaps (CDS). This approach generates a density forecast of severe, nonlin-ear changes in the credit risk premium consistent with the tail risk nature of sovereign distress within a flexible func-tional form. Once a level of credit risk premium under stress is chosen, market- consistent valuation haircuts can be de-rived from standard bond pricing models. Compared to the approaches using changes in government bond yields, sovereign CDS spreads, when available, provide a “pure” mea-sure of maturity- consistent default risk without potential contamination from varying security characteristics and pol-icy measures influencing government bond prices (Box 9.2).

As part of this approach, the determination of a market- implied valuation haircut provides the conceptual founda-tion for incorporating broader bank- sovereign linkages. A higher sovereign risk may also imply a lower probability as well as diminished capacity of governments to bail out banks in a systemic event. These indirect effects may have addi-tional costs ex ante (for example, higher funding costs, espe-cially if the sovereign risk is originally triggered by

deteriorating bank solvency) and ex post (for example, a higher probability of bank failure and lower recovery rates). The effects are usually covered as contingent liabilities in the public debt sustainability analysis of the IMF’s Article IV sur-veillance. When sovereign loan exposures are sizeable, credit risk parameters that are consistent with the market valuation haircut could be more useful to model tail risk when the ac-tual history of the credit risk parameters does not include extreme events.

The chapter is structured as follows. The next four sec-tions describe the key steps of stress testing sovereign risk, which inform the specification of our approach, followed by its empirical application during the European sovereign debt crisis. The final section concludes by summarizing the key aspects of measuring sovereign risk in bank stress tests and providing suggestions for incorporating sovereign risk within integrated stress testing frameworks that model dy-namic and systemic effects from the interaction of credit, market, and liquidity risks.

2. SCOPE OF THE TEST

Exposures and Transmission Channels

The relevant forms of sovereign distress, types of exposures, and channels of transmission differ substantially across countries. For instance, AEs have rarely experienced “crude forms” of sovereign distress such as restructuring and (hy-per) inflation in the post- World War II period, while such incidents have been more frequent among EMDEs (BCBS 2017a). Banks’ sovereign exposures are largely securities in countries with developed sovereign bond markets (AEs and major emerging market countries). When financial markets are underdeveloped (many EMDEs), loans and other types of exposures become more important. It is important to ad-just the scope and the design of sovereign risk stress tests ac-cording to the ecosystem of each financial system.

Box 9.1. Definition of Bank Sovereign Exposures: Basel III Monitoring Exercise

As part of the semiannual Basel III Monitoring Exercise, the Basel Committee on Banking Supervision (BCBS 2018a, 2018b) monitors the exposures of major banks across all member jurisdictions. Sovereign exposures1 are one of the key elements of this exercise and are de-fined as:

• Direct sovereign exposures are exposures to sovereigns (as immediate counterparties). They include both banking book (for example, loans and receivables) and trading book assets (securities and financial instruments, including derivatives and valuation margins). For the monitoring of Basel III liquidity ratios, liabilities from sovereigns are also monitored, including deposits, secured borrowings, derivatives liabilities, and valuation margins, among others.

• Indirect exposures are exposures to counterparties other than the sovereign itself, which are (1) protected (guaranteed) by a sovereign entity, and (2) collateralized by instruments issued by sovereign entities and not subject to haircuts. An example of the latter is a re-verse repo transaction, where a bank swaps an asset for government bond as collateral. Another example is a credit default swap on sovereign securities.

1 “Sovereign” includes a central bank, a central government, multilateral development banks and some international organizations, subna-tional governments, and public sector entities.


Sovereign Risk in Macroprudential Solvency Stress Testing186

Box 9.2. Reasons for Choosing CDS Spread Dynamics for Estimating Valuation Haircuts

A closed- form pricing approach for estimating market- implied sovereign risk using credit default swap (CDS) spread dynamics seems to be preferable to other methods that calibrate sovereign valuation haircuts based on the price volatility of government bonds:

• Risk measurement: CDS spreads are relatively “pure” measures of default risk (IMF 2013a), which might otherwise be “contami-nated” by the price impact of security characteristics (such as coupon frequency, creditor rights, and redemption features) as well as inflation and term premia (and their volatility) if it were extracted from government bond prices. Using CDS spreads also avoids po-tential basis risk from the choice of the appropriate risk- free rate and its term structure impacting the extraction of the credit spread component of government bond yields. In the event of a default, the CDS contract payout usually recovers the par value, which means there is no need to determine the implied default probability (since the recovery rate is endogenized in the observed bond price). Moreover, CDS spreads represent sovereign risk more accurately than sovereign bond yields when yields are kept artificially low by central bank bond purchase programs. However, sovereign CDS spreads could also be influenced by price distortions. Since sovereign CDS contracts for most countries tend to be denominated in US dollars, foreign exchange rate changes (which are often positively correlated with shocks to sovereign risk) could amplify the CDS spread dynamics and lead to a potential overestimation of sovereign risk during times of stress (relative to the dynamics of credit spreads implied by price changes of local currency- denominated government bonds). In addition, CDS contracts provide protection sellers with a “delivery option” (that is, the cheapest- to- deliver government bond), which might raise the credit spread if it implies a relative reduction of the expected recovery rate (relative to that of cash instruments).1

• Market expectations: In addition, the model specification incorporates market expectations of future changes in sovereign risk (as reflected in forward CDS contracts), and, therefore, ensures time consistency between the market- based valuation haircut and the actual valuation change in each year of the stress test horizon. The performance of using forward CDS contracts can also be exam-ined. Appendix Figure 9.2.2 shows the empirical distribution of spot and forward sovereign CDS spreads (with different starting times) for major advanced economies as well as emerging market and developing economies as of the end of 2010 (fitted to the generalized extreme value distribution specified in equation A9.2.24 in Appendix 9.2). These data were used for the estimation of market- based valuation haircuts in Appendix Tables 9.5.1 and 9.5.2. Forward CDS contracts were found to overstate sovereign de-fault risk in the wake of the European sovereign debt crisis but adequately project the potential escalation of sovereign risk in vulner-able countries.2

• Model flexibility and price consistency of shocks: The functional form supports a more nuanced assessment of sovereign risk over the projection horizon and generates tractable estimates of tail events (outside the historical experience, which can be recon-ciled to the probabilistic severity of the overall scenario). This also offers the opportunity to cross- validate other approaches. More-over, the estimated default risk is integrated into an asset pricing model, and, thus, controls for the marginal effect of changes in default risk on the convexity of government bond prices.

1All CDS spreads are derived from over- the- counter markets and tend to be liquid only for a few maturities (compared to government bonds, which, at least in advanced economies, are traded in very liquid markets with wide investor participation). The methodology in this chapter is focused on five- year sovereign CDS contracts, which is the most liquid maturity term (see Tables 9.4 and 9.5).2For most countries, the estimated CDS spread under the two adverse scenarios (defined as the historical density forecast at the 75th and 90th percentiles) exceeded the realized CDS spread— measured at the end of each year (shown as gray dots in the boxplots on the left side of each country chart) and during each year (shown as boxplots on the right side of each country chart), except for the first year of the stress test horizon. Germany, Japan, and the United States notably benefitted from safe haven flows during the European sovereign debt crisis, which resulted in a gradual decline of sovereign CDS spreads. In contrast, for Italy and France, the actual sovereign CDS spreads during the first year of the risk horizon were higher than projected in the mild adverse scenario (75th percentile)—at the 86th and 81th percentiles of the empirical distribution of one- year forward CDS and the 89th and 87th percentiles of the empirical distribution of spot CDS.

includes sovereign securities diminishes. Such liquidity stress could eventually lead to higher overall funding costs.

Economies with Underdeveloped Financial Markets with Higher Outright Sovereign Default Risk

These economies face higher chances of sovereign distress with various forms of default. Delayed interest payments or unilateral debt restructuring— which often constitute “credit events” for CDS contracts (see Appendix Box 9.2.1)—could occur frequently. Such “defaults” could also manifest as the monetization of public deficits and lead to hyperinflation, which would reduce public debt in real terms and weaken the exchange rate (potentially resulting in a currency crisis).

Sovereign distress could be closely related to external vulner-abilities, raising the role of global investors and macro- financial

Economies with Developed Financial Markets with Low Outright Sovereign Default Risk

In these systems, sovereign distress propagates through the valuation shock to sovereign bonds with significant indirect effects. Then a test could focus on securities exposures and apply a valuation haircut to them. This approach implicitly defines sovereign default as a market risk, rather than a credit risk.

A severe indirect channel could stem from the interaction of bank solvency and liquidity. The resulting decline of bank solvency ratio could increase the counterparty risk of the af-fected banks, raising their funding costs, especially when they rely on wholesale funding that is more sensitive to coun-terparty risk than deposits. Banks could struggle to satisfy liquidity requirements, as the value of liquid asset buffer that



conditions. Sovereign risk in many EMDEs tends to be more influenced by external factors than those of AEs. Especially small open economies are susceptible to global demand shocks. Also, external and public sector balances of com-modity exporters could experience large swings along the commodity price cycle. In countries where governments, banks, and nonfinancial firms depend on external finance, global market sell- off events could reduce or reverse capital outflows, resulting in potentially extreme exchange rate and asset valuation shocks in line with a high “beta” of EMDE securities found in empirical studies (IMF 2014a).

Broader types of sovereign exposures become relevant if the state has a considerable role in the financial sector. The prevalence of state- owned banks could create strong cyclical linkages between bank performance and public finance (as well as contingent liabilities). These linkages manifest in in-terest rate controls, directed credit, or financial repression, which may force banks to take on higher credit risk. They may also raise the resolution cost of failed banks.

Determining the Scope

A comprehensive assessment includes all types of relevant sovereign exposures, beyond the valuation of traded expo-sures during times of stress (see Box 9.1).3 In most FSAPs for AEs, solvency stress was mostly driven by the market

3 Given the empirical application of this approach for mostly European countries, it is important to note that many European FSAPs covered all relevant exposures to match the coverage used in EU- wide exercises, which provided the basis for the current BCBS definition (see Tables 9.1 and 9.2). These exercises are the 2011 capital exercise and subsequent tests (EBA 2011a, 2011b, 2012), which covered both direct and indirect exposures similar to the BCBS definition in Box 9.1.

valuation losses from government debt securities, and cash balances at central banks as well as repurchase agreements (repos) or asset swaps were often excluded. Loan exposures are included, but they tend to be a small part of bank assets, and the estimated losses are usually negligible given the lim-ited history of outright sovereign defaults in most AEs (see the following section on loss estimation for details). How-ever, in a macroprudential stress testing exercise for the econo -mies with higher outright default risk and underdeveloped capital markets, it will be essential to think beyond the mar-ket risk aspect, securities exposures, and central government debt, since a larger share of losses is likely to come from loan or loan guarantee exposures to broader government (in-cluding state- owned enterprises). Where needed, a reliable test may require additional data collection to supplement standard reporting.

3. METHOD TO ESTIMATE POTENTIAL LOSSES

Benchmark Approach

Like any other risk factors, sovereign risks generate both ex-pected and unexpected losses impacting bank solvency (see Figure 9.2). Expected losses represent average losses that are likely to materialize in the future based on current informa-tion. These losses affect the capital adequacy ratio (CAR) (that is, capital divided by risk- weighted assets [RWA]) through its numerator— either as a direct hit to capital or through profit and losses), depending on the types of expo-sures (for example, securities held for trading [HfT], avail-able for sale [AfS], and held to maturity [HtM]). In contrast, unexpected losses are extreme losses that tend to occur with a

0

Time

Probability

Aggr

egat

e Lo

sses


(e.g., one standard

deviation)



Expected Loss

Unexpected Loss

Extreme Loss


Provisions

Capital RequirementValue at Risk (VaR)

Conditional Tail Expectation(CTE)

Average

VaR99.9%

CTE99.9%

T

Source: Adapted from Jobst, Ong, and Schmieder 2013.1For loan exposures, the Basel Accord requires banks to set aside loan-loss provisions equivalent to expected losses. Additional provisions (that is, credit cost) in a given year will reduce bank profit and therefore the numerator of the solvency ratio.

Figure 9.2 Conceptual Difference between Expected and Unexpected Losses (Example for loan exposures)1



very low probability— say once in 1,000 years shown as the value at risk (VaR) at the 99.9th percentile in Figure 9.2. These tail risks affect the CAR through its denominator by increasing the capital intensity of assets (that is, risk weights).

Generally, the methods to estimate expected loss differ depending on whether the exposures are in the banking book or trading book (see Table 9.3):

• Trading book exposures are mostly bonds and other market instruments; their expected losses stem from the securities market’s valuation changes.

• Banking book exposures are mostly loans (see Appen-dix 9.2). Therefore, their expected losses are esti-mated with a credit risk approach that includes an empirical satellite model that forecasts credit risk parameters with macro- financial covariates. If banks use internal- ratings- based (IRB) approaches, ex-pected losses are estimated as the product of the probability of default (PD) and loss given default (LGD). Under the standardized approach, the losses are estimated using loan classification (or nonper-forming loan information) or credit rating, assuming a certain level of required provision rate for each category/rating of loans.

Stress tests broadly follow the Basel regulatory capital rules to estimate expected losses but tend to widen the ap-plication of market- consistent valuation. The objective of capital rules is to provide a fair and timely measure of bank solvency without introducing short- term volatility. For in-stance, if banks hold substantial amounts of traded securi-ties, mark- to- market (MtM) valuation changes could lead to frequent changes in regulatory capital, which complicates both lending and investment activities. Therefore, the regu-latory framework (and underlying accounting standards) in-cludes features to smooth out excessive volatility and cyclical effects. In contrast, the main objective of macroprudential stress tests is to examine banks’ resilience to tail events rather than normal cyclical downturns. For this purpose, it is criti-cal to reflect all potential losses immediately and transpar-ently using MtM valuation (that is, the economic valuation approach).4

In doing so, all traded exposures are ideally valued using a market- consistent approach to make stress test results more comparable across banks. The same sovereign securities could be valued differently depending on their accounting treatment (and the way this informs the calculation of the CAR, see Table 9.3). If the market value of sovereign securi-ties declines sharply, it is fully reflected in the valuation of HfT and AfS securities but not necessarily HtM securities, which are valued at amortized cost using historical estimates of credit risk parameters. Some jurisdictions, including Japan and the United States, allow some banks to continue applying the “AfS filter” that limits the impact of short- term

4 Similarly, for banking book exposures, the Basel rules apply cyclically smoothed credit risk parameters called through- the- cycle PDs and LGD; however, stress tests usually apply point- in- time PDs and down-turn LGD as “raw parameters.”

volatility of the value of AfS securities on solvency ratio, though many (European) jurisdictions entirely removed the filter in the mid- 2010s.5 Thus, banks with precisely the same portfolio and balance sheet may have different CARs de-pending on the share of AfS and HtM securities.

The market- consistent approach is also useful for deter-mining expected losses from banking book exposures when a country’s history does not include any sovereign distress episode(s). No empirical satellite model can capture sover-eign distress well without historical precedents. Market val-uation, in contrast, is more sensitive to investors’ perceptions about the likelihood of sovereign distress.

The market- consistent approach is critical when potential regulatory arbitrage or forbearance is a concern. During the European sovereign debt crisis, banks received a one- time supervisory approval to reclassify sovereign HfT and AfS se-curities as HtM (Acharya 2018). While such a measure is vital as a crisismanagement measure that limits the undesir-able amplification effects from the banking sector, it reduces transparency for stress testing. Moreover, there is evidence that banks optimize the accounting treatment of govern-ment debt securities to reduce their capital impact.6

This approach also helps assess the impact of sovereign- bank linkages on bank funding costs. As discussed in Sec-tion 2, one of the sovereign- bank linkage channels is through bank (wholesale) funding costs.7 Investors are likely to pay attention to bank solvency based on a full market valuation in addition to regulatory ratios. Therefore, one approach is to use a sensitivity test that estimates the impact of sovereign distress on market- value- based solvency ratio and then esti-mate its impact on bank funding. The resulting reduction of net interest income could be part of the broader scenario tests where valuation losses from HtM securities are excluded.

However, for unexpected losses from sovereign exposures, stress tests usually follow the regulatory practice, even though it is considered problematic in the financial stability community. Under Basel regulations, local currency- denominated sovereign debt preserves their nominal value during times of stress and thus could be considered “safe assets.” In many cases, these sovereign exposures are as-signed a zero percent credit risk weight under the standard-ized approach (SA) and very low risk weights under the IRB approach if banks estimate PDs and LGDs for sovereigns with no (or limited) distress episodes in the past. While these practices underestimate sovereign risk (Hannoun 2011), recent BCBS regulatory reform efforts to change

5 In the United States, the Federal Deposit Insurance Corporation allowed banks regulated with the standardized approach and savings and loan hold-ing companies to elect a one- time, permanent opt- out from the recognition of unrealized accumulated other comprehensive income (see FDIC 2015). In Japan, regional banks can continue using the AfS filter.

6 For instance, Fuster and Vickery (2018) found that US banks responded to the removal of the AfS filter by reclassifying securities to HtM accounts instead of reducing the portfolio risk of the trading book.

7 Wong and Hui (2009) examine the feedback effects between liquidity and solvency risks. Schmitz, Sigmund, and Valderrama (2017) examine the empirical validity of these feedback effects.



with which the forward yield curve can be estimated. For ad-verse scenarios, the size of shocks may not be sufficiently se-vere in macroeconomic models, which tend to be focused on changes in the short- term (policy) rate (and a projected long- term yield if available). Standard empirical and dynamic sto-chastic general equilibrium models do not endogenously model shocks to financial risk. Also, most of these models do not integrate essential nonlinear effects of financial risks.

Therefore, stress testers often use the market- implied val-uation approach as an alternative statistical method to cali-brate sovereign shocks to sovereign securities. Many FSAPs have used the approach since the European sovereign debt crisis to estimate the haircut to sovereign securities (see Tables 9.1 and 9.2). Section 5 describes a specific modeling technique, which is well suited for the estimation of valua-tion haircuts that capture the tail risk of sovereign exposures under this approach. The haircuts can be derived from the expected change in the price of government bonds in re-sponse to changes in default risk. The price of government bonds broadly reflects two components— the risk- free inter-est rate and the credit risk premium.9 The risk- free rate represents the intertemporal cost of money in line with ex-pected inflation expectations and the real interest rate. The credit risk component signifies sovereign default risk. In the absence of sovereign distress, government bonds are consid-ered safe and yield a risk- free rate of return whose volatility determines any valuation changes. However, when investors recognize higher potential sovereign default risk (or their risk aversion increases),10 they demand a (higher) credit risk premium, which reduces the price of government bonds. Thus, haircuts reflect the differential price impact of higher sovereign risk, with general macro models providing the risk- free component.11 These valuation haircuts are applied to all sovereign exposures for a fully market- consistent capi-tal assessment of sovereign risk; however, empirical con-straints might preclude reasonable estimates for the valuation changes for all types of credit exposures under stress, limit-ing the market valuation approach to capital market instru-ments (in the trading book) only.

For sovereign exposures in the banking book, especially non- capital market instruments, the credit risk approach may substitute for the market valuation approach to determine sov-ereign default risk, particularly in the absence of reliable mar-ket prices. All banks set aside reserves for expected losses from sovereign exposures in the same manner as commercial and consumer loans, consistent with applicable accounting stan-dards. For credit- sensitive exposures, such as loans and receiv-

9 More specifically, government debt yields are determined by (1) macro-economic factors reflecting the evolving monetary policy stance and inflation credibility (which is reflected in the policy rate and inflation expectations); (2) market factors (such as liquidity risk); and (3) fiscal constraints and borrowing capacity (which determine actual and per-ceived default risk).

10 The market price of risk (based on the relation between asset returns and volatility) is a frequently used metric of risk aversion.

11 The risk- free interest rate term structure is assumed to be independent of credit risk in traditional sovereign bond pricing models.

them have not concluded (BCBS 2017a). The challenge is that there are multiple reform approaches and different op-tions, ranging from concentration- based measures to creditrisk- based capital charges, but it is not straightforward to see which one works the best. In the absence of any clear direc-tion regarding potential changes in the regulatory treatment of sovereign risk, most FSAPs, for example, have not changed this practice. Introducing case- by- case adjustments would also reduce the comparability across different exercises.

FSAP Practice

Most FSAPs follow the benchmark approach, with a varying de-gree of valuation practices (see Tables 9.1 and 9.2). Many exercises during the European sovereign debt crisis for the EU Member States applied market- consistent valuations to all sovereign securi-ties including AfS8 and HtM securities except for France [IMF 2013d] and Spain [IMF 2012]). For HtM securities, this meant applying valuation losses instead of provisions according to their credit risks. In the 2012 Italy FSAP (IMF 2013e), the valuation losses from HtM securities were excluded from a macroscenario test but included in a sensitivity test. For these cases, transparency was deemed most important, especially under various crisismanagement measures that mitigated valuation changes (for ex-ample, the European Central Bank’s quantitative easing) and the forbearance. More recent FSAPs have applied the market valua-tion approach to HtM securities less frequently unless banks re-ported a high share of HtM securities or the share rose noticeably. Most FSAPs applied the credit risk approach for assessing sover-eign risks from loans and receivables using historical credit risk parameters. Some FSAPs are attempting to incorporate the indi-rect effects through funding cost as a part of broader efforts to incorporate solvency- liquidity interactions.

For countries with elevated sovereign risk, FSAP exercises have also included valuation losses that were not fully re-flected in prudential reporting. Since CARs do not fully re-flect the short- term cyclical changes in asset valuation, there could be a gap between the current market valuation of sov-ereign debt securities and their valuation in the last reported statutory accounts. Thus, a test would overestimate CAR without adjusting for the valuation gap. Similar adjustments are critical when there is forbearance to manage a crisis or due to the general weakness of the supervisory framework to handle problem assets. However, if sovereign securities are already priced at historically low levels, it may make sense to apply a smaller- than- otherwise shock (in line with adjusting the adversity of macroeconomic scenarios for stress tests that occur when banks already experience some distress).

4. CALIBRATING SHOCKSSovereign risk shocks are difficult to calibrate with standard macroeconomic models. The baseline scenario usually in-cludes the entire yield curve of (own) government securities,

8 The AfS filter still existed during this time period (early 2010s).



190

TABLE 9.1

Treatment of Sovereign Exposures in Mandatory FSAPs in European Union and European System-wide Stress TestsYear Scope of

Sovereign Exposure

Different Treatment for

Domestic Debt? (Y/N)

Scenarios: Baseline (B), adverse (A)

Timing of Shock: All

front-loaded (F) or over time (T)

Valuation Method

HtM AfS HfTIMF FSAP

European Union S-29 Countries

First FSAP (since 2010)*United Kingdom 2011 Y Y Y Y3 A,B T Zero coupon pricing with cash/forward CDS spreads for country-specific shock1

Germany 2011 Y2 Y2 Y N A,B T Zero coupon pricing with cash/forward CDS spreads for country-specific shock1

France 2013 N Y Y Y4 A,B F Discounted cash-flow pricing with cash CDS spreads for country-specific shockItaly 2013 Y2 Y2 Y Y5 A,B T Discounted cash-flow pricing with cash CDS spreads for country-specific shockNetherlands 2011 Y Y Y N A,B T Discounted cash-flow pricing with cash CDS spreads for country-specific shockSpain 2012 N Y Y N A T Zero coupon pricing with cash/forward CDS spreads for country-specific shock1

Belgium 2013 Y Y Y N A,B T Zero coupon pricing with cash/forward CDS spreads for country-specific shock1

Ireland 2016 N Y Y Y5 A,B T Discounted cash-flow pricing with country-specific shock to bond yieldAustria 2013 Y9 Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldLuxembourg 2011 Y Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldSweden 2011 Y Y Y N A,B T Discounted cash-flow pricing with cash CDS spreads for country-specific shock1

Denmark 2014 Y Y Y N A F8 Discounted cash-flow pricing with country-specific shock to bond yieldFinland 2016 Y Y Y Y5 A F Expected losses based on three-notch downgrade using historical PD and LGDNorway 2015 N Y Y N A,B T Discounted cash-flow pricing with country-specific shock to bond yieldPoland 2013 N N N — — — —European Union 2013 N N N — — — —

Second FSAP (since 2010)*

United Kingdom 2016 Y Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yield9

Germany 2016 N Y Y N A F Discounted cash-flow pricing with country-specific shock to bond yieldNetherlands 2017 N Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldSpain 2017 N Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldBelgium 2018 N Y Y N A,B T Discounted cash-flow pricing with country-specific shock to bond yieldLuxembourg 2017 Y Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldSweden 2017 N Y Y N A F Discounted cash-flow pricing with country-specific shock to bond yieldPoland** 2019 Y Y Y N A F Discounted cash-flow pricing with country-specific shock to bond yieldEuro Area Policies 2018 Y Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yield

(continued)


Andreas A

. Jobst and Hiroko O

ura191

Treatment of Sovereign Exposures in Mandatory FSAPs in European Union and European System-wide Stress TestsYear Scope of

Sovereign Exposure



Scenarios: Baseline (B), adverse (A)



Valuation Method

HtM AfS HfTEuropean Authorities (CEBS-EBA-ECB)

EU Capital Exercise 2011 N6 Y6 Y N A F Discounted cash-flow pricing with cash CDS spreads for country-specific shockEU System-wide Stress Test 2010 N N Y N A,B T Discounted cash-flow pricing with cash CDS spreads for country-specific shockEU System-wide Stress Test 2014 N Y Y7 N A T Discounted cash-flow pricing with country-specific shock to bond yieldEU System-wide Stress Test 2016 N Y Y7 N A F Discounted cash-flow pricing with country-specific shock to bond yieldEU System-wide Stress Test 2018 N Y Y7 N A F Discounted cash-flow pricing with country-specific shock to bond yield

Sources: Authors, EBA 2010, 2011a, 2014, 2016, and 2018; ECB 2011; and IMF FSAP country reports.Note: AfS = available for sale; CDS = credit default swap; CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; ECB = European Central Bank; FSAP = Financial Sector Assess-ment Program; HfT = held for trading; HtM = held to maturity; LGD = loss given default; MtM = mark to market; N = no; n.a. = not available; PD =probability of default; Y = yes.1The haircut model in this chapter (Appendix 9.2) was applied in the FSAPs for Belgium (2013), Germany (2011), Hong Kong SAR (2014), Spain (2012), Sweden (2011), and the United Kingdom (2011); other FSAPs followed similar approaches—with an empirically derived sovereign credit spread shock, using either (1) the historical volatility of CDS spreads, such as in the case of France (2013), Italy (2013), Netherlands (2011), Singapore (2013), and Sweden (2011); or (2) the historical volatility of bond yields, such as in the case of Argentina (2016), Austria (2014), Denmark (2014), Indonesia (2017), Ireland (2016), Japan (2012), Mexico (2016), Norway (2015), South Africa (2015), and Korea (2015) as well as most European countries in the second FSAP.2In the FSAPs for Germany (2011), Japan (2012), and Italy (2013), MtM is applied to HtM securities only in separate sensitivity analysis (unlike UK FSAP’s bottom-up test in 2011).3HtM exposures tend to be assessed using the credit risk approach.4Haircuts are applied only to non-”AAA”-rated debt, and French sovereign exposures (“AAA”-rated) were not subject to a valuation haircut.5Only domestic sovereign exposures were stressed.6MtM is applied to HtM sovereign exposures, and the AfS filter was removed.7Including only direct exposures (indirect exposures were covered in the market risk impact).8A part of the overall sensitivity analysis of the capital impact of credit risk.9Following Longstaff and others 2011.

TABLE 9.1 (continued )



192

TABLE 9.2

Treatment of Sovereign Exposures in Mandatory FSAPs (Excluding European Union Countries)Year

(with link to paper)

Scope of Sovereign Exposure



Scenarios: Baseline (B), Adverse (A)



Valuation Method

HtM AfS HfT

IMF FSAPOther Non-EU S-29 and G20 Countries

First FSAP (since 2010)*United States 2010 N N N Y2 A T Distress dependence with emerging market sovereigns3

Japan 2012 N Y Y N A n.a. Discounted cash-flow pricing with country-specific shock to bond yieldCanada 2014 Y Y Y N Single factor F Discounted cash-flow pricing with country-specific shock to bond yieldSwitzerland 2014 Y Y Y Y4 A T No special test for own sovereign (safe haven); the general increase in credit and market risksChina 2010 N N N — — — —Australia 2012 Y Y Y Y4 A T No special test for own sovereign. A general increase in credit and market risksIndia 2013 N N N — — — —Hong Kong SAR 2014 Y Y Y N A,B T Zero coupon pricing with cash/forward CDS spreads for country-specific shock1

Brazil 2012 N Y Y N Single factor F Valuation losses from sovereign yield changesRussia 2011 N N N — — — —Korea 2015 N Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldSingapore 2013 N Y Y N A F Discounted cash-flow pricing with cash CDS spreads for country-specific shockTurkey 2012 N Y Y N A, single factor T Expected losses from valuation changes (market risk) due to higher sovereign riskMexico 2012 N N N — — — —Norway 2015 Y Y Y N A,B T Discounted cash-flow pricing with country-specific shock to bond yieldArgentina 2016 Y Y Y N A F Discounted cash-flow pricing with country-specific shock to bond yieldIndonesia 2010 N N N — — — —Saudi Arabia 2012 N N N — — — —South Africa 2015 Y Y Y Y5 A F Discounted cash-flow pricing with country-specific shock to bond yield

(continued)


Andreas A


ura193


Treatment of Sovereign Exposures in Mandatory FSAPs (Excluding European Union Countries)Year

(with link to paper)

Scope of Sovereign Exposure



Scenarios: Baseline (B), Adverse (A)



Valuation Method

HtM AfS HfT

IMF FSAPOther Non-EU S-29 and G20 Countries

Second FSAP (since 2010)*United States 2015 N N Y Y Single factor F MtM losses from market price and rate movesJapan 2017 Y Y Y Y A T Haircuts for HfT and AfS own sovereign; credit risk approach for HtM securitiesChina 2017 N N N — — — —India 2017 Y Y Y Y5 A F Discounted cash-flow pricing with country-specific shock to bond yieldBrazil 2018 Y Y Y N A T Valuation losses from sovereign yield changesRussia 2016 N N N — — — Data constraints prevented a full analysis6

Turkey 2017 Y N N N A T Credit risk approach6

Mexico 2016 Y Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldIndonesia 2017 N Y Y N A T Discounted cash-flow pricing with country-specific shock to bond yieldSaudi Arabia 2017 Y Y Y N Single factor, A T No emphasis on sovereign risk; single factor interest rate test on bond valuation/credit risk test

Sources: Authors; and IMF FSAP country reports. Note: AfS = available for sale; CDS = credit default swap; FSAP = Financial Sector Assessment Program; HfT = held for trading; HtM = held to maturity; MtM = mark to market; N = no; n.a. = not available; SOEs = state-owned enterprises; Y = yes.1The haircut model in this paper (Appendix 9.2) was applied in the FSAPs for Belgium (2013), Germany (2011), Hong Kong SAR (2014), Spain (2012), Sweden (2011), and the United Kingdom (2011); other FSAPs fol-lowed similar approaches—with an empirically derived sovereign credit spread shock, using either (1) the historical volatility of CDS spreads, such as in the case of France (2013), Italy (2013), Netherlands (2011), Singapore (2013), and Sweden (2011); or (2) the historical volatility of bond yields, such as in the case of Argentina (2016), Austria (2014), Denmark (2014), Indonesia (2017), Ireland (2016), Japan (2012), Korea (2015), Mexico (2016), Norway (2015), and South Africa (2015) as well as most European countries in the second FSAP since it became mandatory for IMF member countries with systemically important financial systems, such as Belgium (2018), Germany (2016), Luxembourg (2017), Netherlands (2017), Spain (2017), Sweden (2017), and the United Kingdom (2016).2Sovereign debt exposures were tested indirectly via market-implied distress dependence, limited to emerging market debt (so domestic sovereign debt was excluded).3Market-based distress dependence between US financial institutions and emerging market sovereigns measured using Segoviano 2006.4Sovereign risk was not tested explicitly, but a general increase of credit risk from higher expected losses due to European sovereign debt exposures.5Only domestic sovereign exposures were stressed.6Credit risk from SOEs were considered more important than market risk.



194

TABLE 9.3

Asset Valuation Rules and Regulatory Capital Impact*

Balance Sheet Items Accounting Standards (for statutory reporting)

Impact on Bank Profits and Regulatory Capital

Economic Valuation (FSAP Principle)

Trad

ing

Bo

ok

Securities held for trading (Hf T)1

Fair value: MtM (or model-based) gains/losses are reported in P&L and taxed.

P&L: Impact of gains/losses on net operating income (“market valuation approach”)

Fair value: MtM (or NPV) gains/losses for all assets (“market valu-ation approach”)4

P&L: Taxable gains/losses (real-ized/unrealized) applied to net operating income before tax.

Regulatory capital: Non-taxable gains/losses that are not distrib-uted as dividends are fully reflected in economic capital.

Securities available for sale (AfS)1

Fair value• Before post-GFC reforms: gains/losses are ac-

counted only as part of equity in financial state-ments (and are not taxed).

• After post-GFC reforms: gains/losses are reported in P&L as OCI (but not taxed).

P&L and regulatory capital• Basel I and II: Valuation impact through

(limited) capital but national discretion to apply the “AfS filter,” which limits the impact of short-term MtM volatility on capital

• Basel III: Valuation impact through P&L; the “AfS filter” is generally removed (but still permitted by some national authorities)3

Ban

kin

g B

oo

k

Securities held to maturity (HtM)1

Loans

Book value: Historical (amortized) cost: expected loss (based on estimated PD and LGD)2 is provisioned and reduces (taxable) net income.

• IAS 39: Backward-looking PD and LGD to calculate incurred losses

• IFRS 9: Forward-looking PD and LGD to calculate expected losses

P&L: Provisioning for expected losses based on accounting rules (“credit risk approach”)

Regulatory capital: Determination of IRB credit risk weights based on forward-looking (12-month ahead) TTC PD and downturn LGD

Sources: Author categories based on Fuster and Vickery 2018, EBA 2011a, 2017a, 2017b, and BCBS 2015, 2017b, 2017c, 2018b.Note: FSAP = Financial Sector Assessment Program; GFC = global financial crisis; IAS = international accounting standards; IFRS = international financial reporting standards; IRB = inter-nal-ratings-based approach; LGD = loss given default; MtM = mark to market; NPV = net present value; OCI = other comprehensive income; PD = probability of default; P&L = profit and loss statement; TTC = through the cycle.*The same valuation rules apply to all securities (sovereign and others).1The exact category names may differ depending on the local accounting rules used in each jurisdiction. For instance, IFRS 9 does not use this nomenclature. Roughly speaking, HfT corresponds to “held with a trading intent,” AfS corresponds to “fair value reported in other comprehensive income,” and HtM corresponds to “fair value through profit and loss” at amortized cost (see Annex V of EBA 2017a). However, US GAAP continues using these categories (under Accounting Standard Codification [ASC] 320). FSAP stress testing exercises have been (and are likely to continue) using these concepts for communication of stress testing method across broad jurisdictions.2The credit risk parameters (PD and LGD) for the calculation of regulatory capital requirements could differ from those applied in statutory reporting (that is, financial statements) based on prevailing accounting standards.3These national authorities include regulators in Japan and the United States. For instance, the recently completed FSAP for Japan (IMF 2017) found that smaller regional banks that are allowed to apply the AfS filter hold substantially more sovereign securities than larger global banks.4The market valuation approach is at the minimum applied to all (traded) government debt securities irrespective of their accounting classification.



cific credit risk component. The sovereign credit shock can be calibrated based on the historical vola-tility of sovereign credit spreads, which can be de-rived explicitly (via sovereign CDS) or implicitly (via excess spreads over a benchmark government bond yield, such as the J.P. Morgan Emerging Markets Bond Index spreads).16 The data can then be para-metrically modeled to generate point estimates of expected default risk at different maturities for each year of the stress test horizon (after controlling for contemporaneous changes in the general level of in-terest rates, which may influence the pricing of de-fault risk). For an adverse scenario, high sovereign credit spreads away from their historical median could be applied (that is, choosing the spreads at the tail of the historical distribution).17

For loans and other non- capital market exposures in the banking book, the credit risk approach could be more suit-able, particularly in the absence of reliable market prices. For loans, these reserves are called loan loss provisions and typi-cally cover the potential losses from NPLs (that are not re-ceiving interest payments). If banks use IRB approaches to determine their capital requirements for credit risk, provi-sions for expected losses are based on estimated through- the- cycle PDs (or better, point- in- time) and downturn LGDs (see Table 9.3). Since credit risk parameters for sovereign expo-sures are likely to behave distinctively from private loans, a separate credit risk model for sovereign risk, which predicts PDs and LGDs or NPLs of sovereign loans as a function of various macro- financial variables, might be warranted. How-ever, when the country’s history does not include any sover-eign distress episode, such a model may not pick a meaningful level of default risk. Under the credit risk approach, the sover-eign risk shock is modeled as a downgrade scenario, which implies a significant and sudden jump of PDs and LGDs.

The sovereign risk shock would be less severe if countries were already in distress. During times of stress, the current sovereign yield curve already includes a level of default risk, which might already be high by historical standards.

16 CDS spreads represent a “purer” measure of credit risk than government bond yields. In the event of a default, the CDS contract payout recovers the par value, which means there is no need to determine the implied default probability (since the recovery rate is endogenized in the ob-served bond price). Moreover, CDS spreads may represent sovereign risk more accurately than sovereign bond yields when yields are kept artifi-cially low by central bank bond purchase programs (see Box 9.2).

17 The accurate expectation of default risk (and its price impact) would ideally be estimated using forward rates on CDS contracts to account for the variability of expected changes in credit spreads over different risk horizons. In contrast, the European system- wide stress testing exer-cise calibrates default risk based on the historical volatility of country- specific spreads over the interest rate term structure of German government debt (EBA 2018a; Table 9.5). This approach has evolved over time from the initial methodology (EBA 2011a), which centered on country- specific credit spreads being set to “common” (global/regional) interest rates and then scaled by the volatility of each country’s spot CDS spread; this introduced a potentially distorting common spread component into country- specific shocks.

ables, these reserves are called loan loss provisions and typically cover the nonaccrual amount of outstanding balances (which can be proxied via changes in nonperforming loans and write- downs). If banks use IRB approaches to determine their capital requirements for credit risk, provisions for expected losses are based on estimated through- the- cycle PDs (or better, point- in- time) and downturn LGDs (see Table 9.3).12 Since PDs (and nonaccruals) of sovereign exposures might change in ways that are quite different from that of commercial and retail expo-sures, a separate provisioning model for sovereign risk might be warranted.13 Otherwise, the sovereign risk shock is modeled as a downgrade scenario, which implies a significant deterioration of PDs and LGDs, resulting in additional provisions to be held for banking book exposures.

In some cases, sovereign risk shocks also include a “com-mon” (global/regional) interest rate component. For stress tests covering a region (such as the EU system- wide stress test) and smaller, open EMDEs, the change in sovereign bond yields comprises country- specific and common global/regional components. Each component covers adverse changes in the risk- free rate and the sovereign credit risk pre-mium at different points of the interest rate term structure of government bonds (see Box 9.1).

• Common interest rate shock: The total change of sov-ereign yields reflects the changes of the risk- free rate and sovereign default risk across multiple countries if widespread concerns about public debt sustainability cause spillover effects within a region. For many smaller, open EMDEs, the interest rates in large ad-vanced economies (especially the United States) have a substantial influence on domestic sovereign yields. If the common interest rate shock is uniform14 and thus results in a parallel upward shift of the yield curve (that is, it does not affect its curvature), the term structure remains unchanged.15

• Country- specific interest rate shock: The primary driver of the sovereign risk shock is the country- spe-

12 The downturn LGD reflects the losses occurring during a downturn in a business cycle, which can be interpreted in many ways (Altman, Resti, and Sironi 2004). One definition of “downturn conditions” is consis-tent with that of a recession—that is, at least two consecutive quarters of negative growth in real GDP. Often, negative growth is also accom-panied by a negative output gap in an economy (where potential pro-duction exceeds actual demand).

13 If only aggregate nonperforming loan (NPL) data for all sectors are available (and if banks apply the standardized approach for unexpected credit risk losses), the sovereign risk shock is modeled as an NPL shock and should be calibrated separately for sovereign and other exposures.

14 Common interest rate shocks might exacerbate sovereign risk in region-ally fragmented economies, which could be addressed by calibrating country- specific shocks based on weights dependent on the level of con-tagion risk between these economies.

15 For example, the European system- wide stress testing exercise (see Table 9.4) took a similar approach until 2016. A 75- basis- point shock (40 percent increase compared to the latest actual yield) was applied to all euro area government bonds and CDS spreads with 10-year maturi-ties. Then, CDS spreads with all other maturities were assumed to increase by 40 percent.



Further raising the default risk could lead to implausibly severe stress, especially compared to countries with stable interest rates.

5. CALCULATING THE CAPITAL IMPACTThe accounting classification of sovereign exposures deter-mines how their expected losses impact banks’ capital ade-quacy under stress. As Table 9.3 shows, security exposures in HfT, AfS, HtM, and loan exposures affect bank capital dif-ferently. Trading losses from HfT securities are considered realized losses, become a part of net income, and are subject to taxation and dividend payout. Assuming all the AfS filters are removed,18 all unrealized gains and losses from AfS secu-rities also become part of net income. However, the unreal-ized valuation changes are not subject to taxation (and are not usually included in dividend payments).19 Therefore, any decline in valuation will reduce capital one- to- one. Expected credit loss from loan exposures will require additional loan loss reserves (LLRs), which are a part of the (taxable) net in-come. Banks usually pay tax and dividend only when tax-able net income is positive.

If the market valuation approach is used to value securi-ties in HtM, the capital ratio under stress is calculated as follows.

CET1

RWA

CET1

Net income

stress

stress

+1

+1

t

t

t

=

+bbefore sovereign losses

LLR for

−∆ ×−∆

MtM HfT

ssovereign loans

d

−( ) −( )1 1 τ

via proffit and loss

− ∆ × +( )+

MtM AfS HtMmax[valuation ggap LLR ]t t for−

,0 HtM

via capital� ��

� ��

RWAt 1++ ∆

RWARWAt

CET1

RWA

CET1

Net income

stress

stress

+1

+1

t

t

t

=


LLR for

−∆ ×−∆

MtM HfT

ssovereign loans

d

−( ) −( )1 1 τ

via proffit and loss

− ∆ × +( )+

MtM AfS HtMmax[valuation ggap LLR ]t t for−

,0 HtM

via capital� ��

� ��

RWAt 1++ ∆

RWARWAt

(9.1)

where d is the dividend payout ratio, τ is the applicable tax rate, LLR denotes the amount of loan loss reserves, ΔMtM is mark- to- market valuation loss of securities (losses carry a positive sign), and ΔRWA defines the possible change in un-expected losses. Time t is the latest actual value before add-ing stress, and t+1 means after stress. Expected losses from HtM securities affects the capital ratio similarly as the AfS

18 If the AfS filter continues to exist, as in some jurisdictions, then AfS se-curities will also have a valuation gap similar to HtM securities (but the size of the gap is different because the two types of securities are valued differently; see Table 9.3). However, unlike HtM securities, there is no LLR earmarked for the existing valuation gap for AfS securities.

19 The accounting rule was changed to reflect all the valuation changes for AfS securities for transparency purposes. All AfS securities are valued at market prices of the reporting date. However, the rule continues to rec-ognize that these are unrealized losses and gains, in contrast to realized trading gains and losses. Therefore, the valuation change from AfS se-curities is taken out from taxable net income.

securities. One difference is that HtM securities are likely to have a valuation gap at time t, which represents the differ-ence between the amortized cost applied to value HtM secu-rities and their market values. The LLR earmarked for HtM securities can cover a part of the gap, but a positive gap is likely to remain. Then the stressed capital ratio represents both existing and additional losses from stress by including the remaining gap to the equation.

Alternatively, if the credit risk approach is applied to HtM, their expected losses are treated in the same way as those from loans. Banks need to set aside additional loan loss provisions to cover the deterioration of credit quality in the HtM securities in the stress scenario.

CET1

RWA

CET1

Net income

stress

stress

+1

+1

t

t

t

=


LLR for

−∆ ×−∆

MtM HfT

ssovereign loans and HtM

d

−( ) −(1 1 τ ))

− ∆ ×via profit and loss

via capitalMtM AfS� ��

� ��

RWARWA

RWt 1+ ∆AAt

CET1

RWA

CET1

Net income

stress

stress

+1

+1

t

t

t

=


LLR for

−∆ ×−∆

MtM HfT

ssovereign loans and HtM

d

−( ) −(1 1 τ ))

− ∆ ×via profit and loss

via capitalMtM AfS� ��

� ��

RWARWA

RWt 1+ ∆AAt

(9.2)

6. EMPIRICAL APPLICATION: EXAMPLES FROM STRESS TESTS IN FSAPS FOR SELECTED EUROPEAN COUNTRIESThis section illustrates the empirical application of the market- consistent valuation approach for assessing sovereign risk consistent with current FSAP practices in macropruden-tial solvency stress tests. It presents a flexible, closed- form approach to calibrating market- implied haircuts using ex-treme value theory (EVT) to capture the impact of signifi-cant shocks to sovereign risk on bank solvency.

Data Collection and Haircut Estimation

For estimating haircuts, the valuation change of government bonds is modeled using the credit risk premium implied in the cost of protecting against sovereign default risk— sovereign CDS spreads (see Appendix 9.2). Since sovereign credit distress is a rare event, the historical CDS spread dy-namics are fitted to a generalized extreme value (GEV) distri-bution to derive the density forecast of a large, nonlinear change in default risk (see Box 9.2).20 The density forecast is then incorporated into the relevant bond pricing formula or proxies of price- yield sensitivity, such as duration and con-vexity of the bonds. The bond pricing formula combines the default risk premium at different maturities of selected gov-

20 The likelihood of the relevant macroeconomic scenario of the stress test could inform the adequate level of statistical significance of the sover-eign risk shock.



ernment debt securities (“benchmark bonds”) with the appli-cable risk- free rate at the beginning of the estimation period.21 The haircuts for the market- consistent valuation of govern-ment bonds differ by the severity of sovereign risk shocks at different maturity tenors and macroeconomic scenarios. This approach was applied— with a full parametric modeling of the CDS spread dynamics— in the FSAPs for Belgium (2013b and 2013c), Germany (IMF 2011a and 2011d), Spain (IMF 2012), and the United Kingdom (IMF 2011b and 2011c).22 Other FSAPs followed similar approaches using either the his-torical volatility of CDS or bond yields.

This approach generalizes the treatment of sovereign risk when the system- wide stress test of the EU banking sector was introduced (EBA 2010, 2011a; Table 9.4).23,24 The daily data from January 2009 to December 2010 are used for the empirical application of the model, so that the results can easily be compared to those used in the two European exer-cises. This cross- validation helps assess whether the method-ology can be a viable alternative to that used by European authorities using greater model flexibility.

More specifically, four steps were followed for deriving the valuation haircuts:25

• Selecting liquid government bonds at different maturi-ties: For each country, the most liquid fixed- rate local- currency- denominated government debt securities (“benchmark bonds”)26 were selected, and groups of bonds maturing within one year around the desired maturity tenor (“maturity buckets”) were created. The sample of bonds was assumed to be representative of typical maturities of bank sovereign exposures (without knowing actual maturity information).

• Estimating the sovereign credit risk shock: For each identified maturity, daily time series data of the spot and forward sovereign CDS spreads27 were obtained

21 For each sample country, a selection of the most liquid (benchmark) bonds is grouped in maturity buckets of one, three, five, seven, and 10 years, with a discretionary margin of +/-0.5 year. When bond- by- bond data are available, a standard bond valuation formula (available in MS Excel® file “IMF Sovereign Risk Stress Testing Tool.xls,” which can be downloaded from the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit) can be used to calculate bond values as dis-counted future coupon and principal repayments and approximate the valuation haircut.

22 The approach was also applied in FSAPs for several non- European countries, such as Hong Kong SAR (IMF 2014c and 2014d).

23 The valuation haircuts were derived based on the market- based ap-proach suggested in this chapter as part of the reference risk parameters for the market risk parameter component of the stress test.

24 The 2011 EU- wide stress test exercise (EBA 2011a, ECB 2011) involved shocks to sovereign spreads through a mixture of an across- the- board increase in yields of 75 basis points, plus a country- specific effect based on bond price movements preceding the forecasting period.

25 The template files for (1) the conversion of government bond data for the estimation of the valuation haircut (“IMF Sovereign Risk Stress Testing Tool.xls”), and (2) the download of historical CDS spread data (“Data_Input.xlsx”) can be obtained at https://www.elibrary.imf.org/page/stress-test2-toolkit.

26 Since the credit and interest rate assumptions refer to domestic currency yield curves it is necessary to choose local currency debt only.

27 The CDS spreads were calculated consistent with the standard pricing formula using the “fair value model” of the International Swaps and

to estimate the historical spread dynamics and deter-mine the market- implied default rate.28 The recovery rate is endogenized in the default rate implied by the observable spread. The variation of spread changes over a sufficiently long estimation period29 was then calibrated using the GEV distribution. The distribu-tion is suited for modeling tail events and provides a closed- form expression of their asymptotic tail be-havior.30 Point estimates of expected PD at certain levels of severity (for example, percentiles) were then obtained for each year of the five- year test horizon. For the baseline scenario, the last observable current or forward CDS spread (whichever is larger) was chosen to reflect current market expectations. For adverse scenarios, higher country- specific credit shocks at the 75th percentile (and higher) of the fore-casted distribution were applied.31 In all scenarios, the sovereign credit spread shock was estimated with and without a common interest rate shock of 50 ba-sis points.32

• Calculating individual valuation haircuts: The hair-cuts were set as the expected change in the prices of selected benchmark bonds vis- à- vis their market value as of the data cut- off date. The price change corresponds to the total yield changes, including the effects of the expected PDs and the risk- free rate, which varies across maturity tenors. Within each maturity group, individual bonds were priced over a five- year stress test horizon using both the adjusted zero- coupon bond and discounted cash flow meth-ods and considering the specific maturity dates, coupons, and coupon frequencies.33

• Determining the aggregate valuation haircut: The hair-cuts for the individual bonds were then aggregated to country- specific haircuts for each maturity group by taking weighted averages using the outstanding

Derivative Association (see Appendix Box 9.2.2) at quarterly payment frequencies. For euro area countries, only euro- denominated bonds were considered.

28 The reference assets for the forward CDS spreads are the selected sample bonds.

29 In the example, the estimation period is limited to two years (January 1, 2009 through December 30, 2010) since reliable sovereign CDS spreads for advanced economies were available only from January 1, 2009 (see Table 9.4).

30 The cumulative GEV distribution function is calibrated under the up-per bound assumption of both mean and variance being defined (see Appendices 9.2 and 9.3).

31 The selection of the 75th percentile is consistent with the guidance on reference parameters for market risk in the EU- wide stress test in 2011. In the adverse scenario, they were set to the 25th percentile of the em-pirical distribution.

32 The importance of separately modeling the country- specific sovereign spread shock is consistent with recent evidence in Crump, Eusepi, and Moench 2018, which found that term premiums account for the bulk of the cross- sectional and time series variation in yields and largely explain the yield curve’s reaction to structural economic shocks.

33 Since the CDS spread curve flattens significantly beyond the five- year maturity, we focused on a five- year maturity tenor of credit spreads and selected benchmark bonds.







198

TABLE 9.4

Comparison of Sovereign Valuation Haircut Methods in IMF FSAP and EU System-wide Stress Tests IMF FSAP1 EBA-ECB

Since 2010 2011 2014, 2016, and 2018

Calculation of Sovereign Risk Shock2

Common shock (General yield curve)

Upward shift in the yield curve via constant or maturity-dependent increase of risk-free rate (for the zero-coupon bond pricing formula) or a marginal change in the yield-to-maturity (for the discounted cash flow formula)

Upward shift of the weighted average of national sover-eign CDS curves—spread shock consistent with general macroscenario of a 75-basis-point spread increase on 10-year euro area bonds—and application of the relative average increase (that is, 40 percent) proportionately to all other maturities of the CDS curve without altering its shape

Upward shift in the yield curve as general interest rate im-pact via maturity-specific marginal change in the yield-to-maturity using changes of swap rates (discounted cash flow formula)

Data source Market-based; spot and forward CDS spreads Market-based; spot CDS spread Market-based; government bond yields and dependence structure between US and German bond yields

Idiosyncratic shock (Country-specific default risk)

Added to spread increase under common shock (if any) Added to spread increase under common shock Added to spread increase under common shock implicit in general interest rate increase

Measure Based on past spread changes of forward CDS spreads (for each country), estimated separately for each maturity of benchmark bonds and each period of the forecast horizon

Based on past spot CDS spread changes (for each coun-try), estimated separately for each period of the forecast horizon

Implicit in the country-specific interest rate changes (“yield shock”) of government bonds (for each country) via credit spread impact (sovereign spread over swap) for each period of the forecast horizon

Statistical support Current expectations from latest forward CDS spread as well as different percentiles of the parametrically esti-mated density forecast for adverse scenarios

Historical (daily) volatility over the last month preceding the forecast horizon

Historical volatility of CDS spreads; country-specific shocks to EU long-term interest rates capture the spillover impact from the initial US bond yield shock to German long-term yields and the widening of credit spreads im-plied by change in sovereign bond yields of EU countries4

Maturity tenor3 Large part of term structure (1, 3, 5, 7, and 10 years) Almost entire term structure (3 months as well as 1, 2, 3, 5, 10, and 15 years)

Almost entire term structure (3 months as well as 1, 2, 3, 5, 10, and >10 years)

Estimation time period Flexible, limited by data availability only (for example, Jan. 1, 2009, to Oct. 30, 2013 (>4.5 years) in the case of the FSAP for Hong Kong SAR)

Between Oct. 31 and Dec. 1, 2010 (≈ 1 month) Between Aug. 3, 2012, and Dec. 31, 2013 (≈ 1.5 years), for 2014 stress test; n.a. for exercises in 2016 and 2018

Forecast (stress) time horizon

Multiple periods (5 years) Front-loaded shock, single period (1 year) Multiple periods (3 years) in 2014; front-loaded shock, sin-gle period (1 year) for stress tests in 2016 and 2018

(continued)


Andreas A


ura199


Comparison of Sovereign Valuation Haircut Methods in IMF FSAP and EU System-wide Stress Tests Calculation of Valuation Haircut5

Valuation method Zero-coupon bond pricing and discounted cash flow method6

Discounted cash flow method

Bond selection Representative selection of outstanding local currency-denominated sovereign debt in each sample country (“benchmark bonds”); creation of maturity group of bonds maturing within a short time window around the desired maturity tenor

Maturity group 5 years (extendible to any selection of maturity group) 3 months as well as 1, 2, 3, 5, 10, and 15 years 3 months as well as 1, 2, 3, 5, 10, and >10 years

Length of window around maturity groups (“matu-rity buckets”)

±6 months ±2 months (3-month maturity group) to ±3 years (10-year maturity group)

Yield adjustment Adjustment of yields to take into account the change of yields between the end point of the estimation window and starting point of the stress period

Applicable scenarios Baseline and adverse scenarios Adverse scenario only

Scope of application (expected losses)7

Sovereign exposures (direct and indirect) in both trading and banking books assessed under market valuation

approach; possible exception of loans and receivables under credit risk approach8

Sovereign exposures (direct and indirect9), assessed at fair value (available-for-trading (AfT) and available for sale (AfS)) under market valuation approach;10 held-to-maturity (HtM) securities as well as loans and receivables in the

banking book assessed under credit risk approach

Published information Valuation haircuts for all relevant countries, normally published in a “Technical Note on Stress Testing”

together with the “Financial System Stability Assessment” report

Valuation haircuts, by country Valuation haircuts, by country

Sources: Authors’ research; EBA 2010, 2011a, 2014, 2016, and 2018b; ECB 2011; ESRB 2015; and IMF FSAP country reports.Note: EBA = European Banking Authority; European Central Bank; CDS = credit default swap; FSAP = Financial Sector Assessment Program: Y = yes, N = no; n.a. = not available.1The haircut model in this chapter (Appendix 9.2) was applied in the FSAPs for Belgium (2013), Germany (2011), Hong Kong SAR (2014), Spain (2012), and the United Kingdom (2011); other FSAPs followed similar approaches using either the historical volatility of CDS or bond yields.2Changes in country-specific default risk over a specific time horizon. 3Since the credit spread curve tends to flatten beyond the five-year maturity, the extension of default risk shocks over longer maturities produces similar results. 4This method was used only for cross-validation to replicate the EBA-ECB approach using forward CDS spreads. 5Changes in prices of benchmark sovereign bonds over a specific time horizon subject to estimated spread shocks to yield. 6The deviation of US long-term bond yields from the baseline considered in the EBA/ECB Banking Supervision adverse scenario is broadly similar in magnitude and profile to what was used in the adverse scenario of the November 2013 Comprehensive Capital Analysis and Review stress test conducted by the US Federal Reserve. 7In addition, banks were requested to compute (stressed) regulatory risk-weighted assets according to the applicable prudential framework. 8In most FSAP stress tests of major economies since 2011, the banking book has also been fully subjected the market-based valuation haircut with and without the “AfS filter” (see Tables 9.1, 9.3, and 9.4).9Indirect exposures only for sovereign positions in the trading book. 10For sovereign positions in the banking book, banks were requested to estimate impairments/losses in line with sovereign downgrades.



Under a severe adverse scenario, sovereign haircuts on stressed European countries average 15 percent during the first year of the stress test horizon. As of the end of 2010, forward CDS spreads indicate elevated expected default risk relative to the historical experience. Actual CDS- implied de-fault risk of stressed European economies was already much higher than their historical average (and higher than the 75th percentile of the density distribution) at that time. In the case of Greece, forward prices on CDS imply near de-fault, which pushes the haircuts based on actual end of 2010 data beyond the 99th percentile (not reported). The results for other European countries are relatively benign at an aver-age haircut of about 5 percent during the first year of the test horizon. There are few (if any) additional haircuts beyond 2011, given the flattening of the CDS curve at longer ma-turities and heavy discounting of bonds issued by stressed countries during 2011.

7. CONCLUSIONThis chapter presented how to stress test for sovereign risk, largely based on FSAP experiences, with a particular focus on a novel approach for calibrating market- consistent valua-tion haircuts. Macroprudential solvency stress tests, such as those in FSAPs, share the following common characteristics in assessing the capital impact of sovereign distress:

• Comprehensive scope: It is ideal for covering all sover-eign exposures in both the trading and banking books, for instance, by following the BCBS’s semiannual Basel III monitoring exercises (BCBS 2018a and 2018b), including indirect exposures that are ei-ther government- guaranteed or collateralized by in-struments issued by sovereign entities. Nonetheless, the structure of sovereign exposures (and their mate-riality) or data constraints vary across countries and may require narrowing the scope to (1) market valu-ation losses from government securities (mostly for banks in AEs) and (2) higher provisions for loan ex-posures to general government and state- owned en-terprises, which often dominate sovereign exposures of banks in EMDEs.

• Market- consistent valuation: The market valuation approach provides a transparent capital assessment of sovereign risk. Applying this approach to all se-curities, including HtM securities, allows the most transparent and comparable assessment across banks and jurisdictions. The treatment of HtM se-curities varied across FSAPs. In most cases, the credit risk approach was applied to loans and re-ceivables (to capture the impact of impairments and downgrade risk); however, this approach might underestimate potential losses if there is no major distress event in the historical data. In these cases, using credit risk parameters consistent with the market valuation approach can generate sufficiently severe shocks.

amounts of these bonds as weights.34 Since the valu-ation haircut is specific to each benchmark bond, the aggregate valuation haircut for each country repre-sents the weighted- average change in market valua-tion over the relevant stress test horizon. This implies that banks hold portfolios of sovereign debt securi-ties similar to the portfolio of benchmark bonds used for deriving the aggregate valuation haircut when accurate portfolio data are not available (or cannot be accessed for estimation of the valuation haircuts).35 However, if complete portfolio data are available, the valuation haircut can be more nuanced within each maturity bucket based on the term structure of credit risk premiums.36

Findings

The estimated haircuts are consistent with those in the Euro-pean stress testing exercises but provide a more comprehen-sive and nuanced assessment. Table 9.5 provides the estimated valuation haircuts for sovereign exposures with an average maturity of five years in the baseline scenario and two adverse scenarios at the end of 2010 (see also Appendix Tables 9.5.1 and 9.5.2 for detailed results, including for non- European countries). The haircuts are broadly comparable to those used in the first European system- wide stress tests (EBA 2010, 2011a; ECB 2011).37 However, the severity of haircuts seemed more plausible and differentiated across countries due to greater model flexibility regarding statisti-cal confidence and configuration of interest rate shocks (see Box 9.2).38 The distribution- based model specification also considers the market- implied assessment of future changes in sovereign risk and therefore enhances the analysis of sov-ereign risk by anchoring the calibration of shocks in market expectations.

34 The valuation of government bonds and the capital impact of market- implied sovereign haircuts (using historical data) might differ from that in the past due to changes in structural and macroeconomic conditions. For instance, IRB banks might need to increase risk weights for certain government bonds to better reflect their true riskiness. Also, monetary normalization will raise the “market price of risk,” which may also af-fect investor risk appetite to take on sovereign risk. While these changes can make government bond prices more risk- sensitive, they are likely to be immaterial relative to the extreme increase of default risk under stress.

35 Alternatively, one may pick a representative maturity as a general as-sumption of the interest rate elasticity of traded sovereign exposures.

36 If the size- weighted maturity profile of sovereign portfolios is signifi-cantly different from the maturity terms at which valuation haircuts have been estimated, these haircuts may be adjusted to match the actual key rate durations.

37 In the case of Greece, the estimated haircuts are close to those used in the European system- wide stress testing exercise (EBA 2011a), and in line with the average 21 percent mark- down of private creditors (Boone and Ardanga 2011).

38 The approach does not seem to be influenced by liquidity concerns in the sovereign CDS market. Historical data (since January 2011) suggest the pricing and trading volume of sovereign CDS spreads in the most significant sample countries are only weakly correlated.


Andreas A


ura201

TABLE 9.5

Comparison of Sovereign Valuation Haircuts in European Stress Tests (2010-2011) and Results in IMF Stress Testing (Percent)Five-Year Maturity

CEBS (2011)1 EBA (2011)2 IMF (2011)3

Without common shock With common shockBaseline Adverse Adverse Baseline Adverse Baseline Adverse

At 75th pct. At 90th pct. At 75th pct. At 90th pct.2010 2011 2010 2011 2011 2011 2011 2011 2011 2011 2011

Euro area Core Austria 1.0 2.8 3.1 5.6 3.4 2.1 1.3 2.2 4.3 3.5 4.5 Belgium 1.4 3.1 4.3 6.9 5.9 7.2 3.9 7.8 9.5 6.3 10.0 Finland 0.0 3.3 1.9 6.1 2.7 0.9 0.5 0.8 3.2 2.8 3.0 France 1.5 3.0 3.7 6.0 4.1 3.5 2.0 3.7 5.8 4.4 6.1 Germany 0.1 2.5 2.3 4.7 2.1 2.1 0.8 1.4 4.4 3.2 3.8 Netherlands 1.1 2.5 3.0 5.2 3.2 1.8 0.9 1.5 4.0 3.1 3.8 Stressed Greece 3.9 4.3 20.1 23.1 12.6 3.7 11.9 27.0 5.9 13.9 28.7 Italy 1.2 2.9 4.9 7.4 8.4 5.5 4.0 8.0 7.7 6.2 10.1 Ireland 1.6 4.2 8.6 12.8 12.6 12.0 10.9 22.4 14.3 13.3 24.4 Portugal 2.3 3.7 11.1 14.1 11.6 10.4 8.4 17.5 12.5 10.6 19.4 Spain 1.3 4.1 6.7 12.0 9.0 8.7 5.5 11.3 11.1 8.0 13.6 Non-euro area Czech Republic 0.0 2.7 4.6 11.4 3.2 0.8 1.3 2.2 3.2 3.6 4.5 Denmark 0.0 1.4 2.1 5.2 2.6 1.2 0.6 1.0 3.6 3.0 3.4 Norway n.a.4 n.a. n.a. n.a. 1.5 n.a. n.a. n.a. n.a. n.a. n.a. Poland 2.6 6.1 6.4 12.3 2.8 2.3 2.3 4.0 4.7 4.8 6.5 Sweden 1.3 2.3 5.0 6.7 1.9 1.1 0.9 1.6 3.4 3.2 3.8 United Kingdom 5.0 6.9 7.7 10.2 4.7 1.0 1.2 2.2 3.4 3.6 4.6

Sources: CEBS; EBA; ECB; and authors.Note: Estimation based on the valuation haircuts of benchmark bonds at five-year maturity. CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; n.a. = not available; pct = percentile.1Valuation haircuts under both baseline and adverse scenarios (CEBS 2010b).2Valuation haircuts under the adverse scenario (EBA 2011a).3IMF Financial Sector Assessment Program haircuts using the zero-coupon pricing approach based on current market expectations using end-year forward credit-default-swap prices (baseline scenario) as well as the 75th and the 90th percentiles of the empirically fitted density forecast of a country-specific credit spread shock (adverse scenario), with and without a common interest rate shock of 50 basis points for all countries.4No availability of liquid benchmark bonds/CDS swaps for Norway.



historical spread dynamics of spot and forward sover-eign CDS. This approach allowed for deriving the density forecast of severe, nonlinear changes in the credit risk pre-mium consistent with the tail risk nature of sovereign dis-tress within a flexible functional form. CDS spreads, when available, tend to provide a “pure” measure of maturity- consistent default risk than bond yields.

An integrated sovereign risk assessment for macropruden-tial surveillance and financial stability analysis will require additional work. The market valuation approach focuses on the direct impact of sovereign distress on bank solvency but does not consider other transmission channels across sectors and countries. Such feedback effects can be assessed more comprehensively by either (1) interacting sovereign debt sus-tainability analysis and bank stress tests or (2) estimating the effects in empirical multisector models (such as Global Vec-tor Autoregressive approaches), codependence models for both banks and sovereigns, or general equilibrium models with bank and sovereign distress. In addition, the interac-tion between solvency and liquidity conditions under stress could be explicitly addressed as part of integrated stress test-ing frameworks that model dynamic and systemic effects from credit, market, and liquidity risks. For example, the implications of higher sovereign risk on bank profitability and liquidity risk due to higher funding costs could be ex-plored, as well as the implications of setting higher haircuts on government debt as a key component of bank liquidity buffers.39 While these models are being developed, it is still hard to assess their performance.

39 In principle, the same estimation approach could be used for gauging haircuts to liquid assets in liquidity stress tests. The haircut for a liquid-ity stress test should be higher than those used for a solvency test (for example, by taking the shock from tails of the distribution). The time horizon for a liquidity stress test is much shorter than that of a solvency test, and the distribution of yield changes within a month is much wider than the distribution of annual yield changes.

• Unchanged risk weights: Capital requirements for un-expected losses from local sovereign exposures are very low due to their status as “safe assets.” Stress tests typically maintain the prevailing capital inten-sity since the capital impact of revising the risk weights for sovereign exposures is likely to be very large, and policy discussions on reforming the cur-rent regulatory treatment are evolving.

• Adjusting for existing losses for sovereigns with ongoing distress: When stress is already ongoing, the latest market valuation could be even lower than the value reflected in solvency ratio for some exposures. Then, it is more transparent to separate deterioration of sol-vency ratio due to already materialized stress from additional stress in the adverse scenario.

• Integrating sovereign risk into the macroeconomic sce-nario: Where there are higher chances of outright sovereign default in economies where a large part of sovereign exposures are loans and guarantees (in-cluding state- owned enterprises), a more extensive range of macro- financial spillover effects becomes more important. Then, focusing on the valuation changes with sovereign securities may become too narrow. A more comprehensive approach, including an effort to embed them in a macroscenario— the monetization of fiscal deficits (or large fiscal deficits with loose monetary policy) and resulting hyperin-flation and currency crises with capital outflows— is likely to be essential.

When calibrating the valuation haircuts for sovereign se-curities, the chapter’s approach underscores the importance of accounting for the tail- risk nature of sovereign risk. The potential losses from sovereign risk are likely to have a long tail: there is a very small chance that could cause extreme losses. Without using an adequate method, a stress test is likely to underestimate the potential impact. The chapter presented the method that fits a GEV distribution to the


Appendix 9.1.Interaction and Feedback between the Sovereign and Financial Sector Balance Sheets Using Contingent

Claims Analysis

Contingent claims analysis (CCA)40 is used to illustrate the interaction between the sovereign and financial sector balance sheets and the potential rise of their respective credit spreads during stress episodes (Gray and Jobst 2010a, 2010b). In the fol-lowing analysis, it was assumed that banks are the only relevant financial institutions for the assessment contingent liabilities from this interaction.

The expected losses from total sovereign debt of a country can be expressed as a European put option, where the underlying asset is government asset A, the strike price is debt amount D, and maturity is sovereign debt maturity T, so that

t t sov sov

r T tL P A D t T x D et T

( ) ( , , , ) ( ),

(+ −

− −= = −τ )) ( ) ,− − + x Asovt�

(A9.1.1)

where (·) is the cumulative distribution function of the standard normal distribution, with

xT t

A

Dr

A

sov

sov

A

sov

t

t T

sov± =

−+ ±

12

2

σσ

ln�

,

(( ) .T t−

(A9.1.2)

The term structure of the corresponding sovereign credit spread ssovt can be used to (1) estimate the implied value of sover-

eign assets Asovt� and asset volatility σAsov

2 and (2) calibrate a risk- adjusted measure of market- implied sovereign risk (using the sovereign balance sheet) in the absence of measurable equity and equity volatility for sovereign debtors. Sovereign spreads (in basis points) are defined as

sT t

lnP A D t TD esov

sov

sovr T tt

t T

= −−

−− −

11

( , , , )

,( )

×10 000,

(A9.1.3)

with sovereign default barrier, Dsovt T, (or threshold that debt restructuring is triggered), over time horizon T−t at risk- free dis-

count rate r, subject to the duration of debt claims, the leverage of the firm, and asset volatility. Rearranging the first equation above for the implicit sovereign put option gives

Psov

sovr T t

sovA D t TD e

xA

Dt T

t( , , , )( )

,( )− − −= − −

�

ssovr T t

t Te

x,

( )( )

− − +−

(A9.1.4)

so that

sT t

ln xA

D esovsov

sovr T tt

t

t T

= −−

− − −− − −

11 ( )

,( )

� ( ) , .−

×+x 10 000

(A9.1.5)

40 The CCA is a generalization of option pricing theory pioneered by Black and Scholes (1973) and Merton (1973, 1974), which stipulates that equity can be modeled as an implicit call option, while risky debt can be modeled as the default- free value of debt less an implicit put option that captures expected losses.



The sovereign default barrier (based on available information on the periodic debt service) and the observed sovereign credit spread at the weighted average maturity of the debt repayment schedule can be used to estimate the implied sovereign asset value, which is defined as

A R PVPS P A D t T Othersov t t bankt

( , , , ) ,= + + α +

(A9.1.6)

comprising (1) foreign currency reserves, R; (2) the present value of the primary fiscal surplus (or net fiscal assets), PVPS; (3) the implicit and explicit contingent liabilities from the aggregate banking sector risk, P A D t Tbankα ( , , , ) ; and (4) remainder items (“Other”). The contingent liabilities are defined as the share α of expected losses in the banking sector, which are defined—analogous to expected losses from sovereign risk— as put option

P A D t T x D e x Abank bank

r T tbankt T t

( , , , ) ( ) ( ) .( ),

�= − − −−− −

+ (A9.1.7)

Since the contingent liabilities can be estimated using the Systemic CCA framework (Gray and Jobst 2011a, 2011b),41 and the value of reserves and the primary fiscal balance are observable, the above equation is solved for the residual (“Other”). “Other” includes a number of public sector assets and various unrealized liabilities, such as pension and healthcare obligations as well as contingent financial support to nonbank financial institutions, guarantees from other governments or multilaterals, and/or backstop assets (for example, land or other public sector assets). Thus, this valuation approach helps assess the effect of changes in any constituent component of sovereign default risk—reserves, the primary fiscal balance, and the implicit banking sector guarantee—on the sovereign asset value (and corresponding sovereign credit spreads) for sensitivity analysis and stress testing.

Conversely, the effect of contingent liabilities on the credit spreads of banks is a function of the implicit put option, P A D t Tbank ( , , , )α (derived from equity information), times the fraction of risk − α1 retained by banks plus a premium δ( ) if

high sovereign spreads spill over to increase bank spreads such that

ST t

lnP A D t T

Dbankbank

bankt

t T

= −−

− −11

1( ) ( , , , )

,

αee r T t− −

+

×

( ), .δ 10 000

(A9.1.8)

This simple model shows how sovereign and bank credit spreads can interact and potentially lead to a destabilization pro-cess. Higher sovereign spreads can cause higher bank spreads as (1) the value of the implicit bank put option for sovereign guarantees decreases (that is, α declines); (2) the value of the bank’s holdings of government debt decreases; and (3) the bank default barrier may increase due to higher borrowing costs as the premium (δ) increases.

41 The Systemic CCA framework was applied in the macroprudential stress tests of banking sectors as part of the IMF FSAPs for Germany, Hong Kong SAR, Spain, Sweden, the United Kingdom, and the United States.


Appendix 9.2.Estimating Haircuts for Sovereign Risk

Sovereign valuation haircuts can be derived from the expected change in the price of government bonds consistent with the estimated change in market- implied sovereign default risk. The haircuts differ by the severity of shocks to sovereign risk at dif-ferent maturity tenors and macroeconomic scenarios. The sovereign risk shock is modeled using forward- looking information from past changes in the cost of sovereign default risk protection.

The estimation draws on different data sources (see Appendix Figure 9.2.1). For each country, the most liquid fixed- rate, local- currency- denominated government debt securities (“benchmark bonds”)42 with residual maturity up to 10 years are se-lected, and groups of bonds maturing within one year around the desired maturity tenor (“maturity buckets”) are created. The valuation change of these bonds under a particular scenario is calculated by combining the default risk premium at different maturities with the applicable risk- free rate at the beginning of the estimation period (which is equivalent to the valuation haircut relative to the prevailing market value of each bond).43 The default risk premium compensates for the expected default

42 Since the credit and interest rate assumptions refer to domestic currency yield curves, it is necessary to choose local currency debt only.43 For each sample country, a selection of the most liquid (benchmark) bonds is grouped in maturity buckets of one, three, five, seven, and 10 years, with a

discretionary margin of +/−0.5 year. When bond- by- bond data are available, a standard bond valuation formula (see MS Excel® file “IMF FSAP Sover-eign Risk Stress Testing Tool.xls” available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit) can be used to calculate bond values as discounted future coupon and principal repayments to approximate the valuation haircut.

Source: Authors.

Appendix Figure 9.2.1 Overview of Haircut Valuation Methodology for Sovereign Exposures ( Five- year stress testing horizon)

Forecast horizon

Year5

Year4

Year3

Year2

Year1

Size of commonspread shock

Country-specificshock

(Baseline/adverse

scenarios)

Determination of risk-free rate(At starting point of forecast)

Commonshock

(Baseline)

Estimated haircut

DATA INPUTFor each country

CALCULATION AND ESTIMATION

1

2

3

Sovereign CDS—forward contracts—

With maturities matched to those of sovereign

bonds (see below)

Sovereign CDS—spot/cash—

With maturities matched to those of sovereign

bonds (see below)

Sovereign(benchmark) bonds

Grouped in “maturity buckets” (1, 3, 5, 7, and

10 years)

Observed forward spread

Annual averageOR

Last observation in estimation window

For each maturity

Observed cash spread

Annual averageOR


Observed yield to maturity


Density forecast of forward spread

(75th and 90th percentiles)

Based on parametric fit

For each maturity and each year of

the forecast horizon




risk implied by the historical spread volatility of sovereign credit default swaps (CDS). The spread dynamics inform the density distribution of expected default risk over the stress test horizon.44 This approach generalizes the treatment of sovereign risk in the EU- wide stress testing for banks in an integrated asset pricing framework using the price dynamics of CDS rather than government bonds to calibrate a market- consistent sovereign risk shock (see Table 9.5).45,46

Since the valuation haircut is specific to each benchmark bond, the aggregate valuation haircut for each country represents the weighted- average change in market valuation over the relevant stress test horizon. In the application of these haircuts to sovereign exposures, banks are assumed to hold portfolios of sovereign debt securities similar to their supply when accurate portfolio data are not available.47

SPECIFICATION OF THE RISK-FREE RATE AND THE CREDIT RISK PREMIUMFirst, the prevailing risk- free rate is determined, and the credit risk premium under baseline conditions is specified. The standard pricing formula for a coupon- bearing bond is reconciled with the zero- coupon bond pricing formula (assuming equivalence of economic value) to project future bond prices contingent on changes in idiosyncratic risk (with the possibility of considering a general shock to interest rates). This is done for selected outstanding (fixed rate) bonds b( )1 of each sample country j J∈ , which are grouped by residual maturities y k k[ 0.5, 0.5]∈ − + in predefined “maturity term buckets” of k K {1,3,5,7,10}∈ = years.

Since each sample bond carries regular coupon payments, c, with a payout frequency m in each year n, the observed market price Pb j k t, [ ],1

conforms to the discounted cash flow (DCF) pricing formula

Pc

rp

rb j k t nT t n

b j k tn m1

1

1 1 1, [ ],, [ ],

/( ) (=

++

+=−∏

bb j k tT t b j k tP

1

2, [ ],

, [ ],) −≥

(A9.2.1)

with yield- to- maturity (YTM) rb j k t, [ ],1 at time t (which determines the “data cut- off” for the estimation window over the re-

maining life of the bond T−t), the notional amount (or principal) p, and exceeds the zero coupon bond price Pb j k t, [ ],2 by con-

struction since → =r r

c b j k t b j k tlimo , [ ], , [ ],2 1

. Pb j k t, [ ],1

can be transformed into the (nonobservable) equivalent of a zero- coupon bond price (“zero-coupon equivalent,” or ZCE)

P Pc

rb j k tZCE

b j k t nT n

b j1 1

1

11

1, [ ], , [ ],, [(

= −+=

−∏kk t

n m tb j k t

T t t b j k

pr

P],

/, [ ],

, [ ]) ( )+ =

++ =

−ϕ ϕ

11

2 ,,t

(A9.2.2)

by stripping away all coupon payments c (with payout frequency m in each year n)48 and adjusting for the first- and second-order pricing effects of the missing coupon payments (that is, the positive impact of removing coupon payments on the price sensitivity of the bond relative to a lengthening of the duration), with adjustment factor

( )12

( ),, [ ],

, [ ], , [ ], , [ ],

, [ ], , [ ], , [ ],21

1 2 1

1 2 1

PD r r

C r rt b j k t

b j k tcoupon

b j k t b j k t

b j k tcoupon

b j k t b j k t

ϕ =−

+ −

(A9.2.3)

where the marginal duration and convexity49 attributable to the coupon payment are

∏=+ +=

−Dr

ncm rb j k t

coupon

b j k tnT t n

b j k tn m

1(1 ) (1 ), [ ],

, [ ],1

, [ ],1

1 1

(A9.2.4)

and

∏= −

+=−C D

crb j k t

couponb j k tcoupon

nT t n

b j k tn m(1 ), [ ], , [ ], 1

, [ ],1 1

1

(A9.2.5)

44 The assumption is based on the empirically derived probability function of forward rates of sovereign CDS contracts.45 The valuation haircuts were derived based on the market- based approach suggested in this chapter as part of the reference risk parameters for the market

risk parameter component of the stress test.46 The 2011 EU- wide stress test exercise (EBA 2011a, ECB 2011) involved shocks to sovereign spreads through a mixture of an across- the- board increase in

yields of 75 basis points, plus a country- specific effect based on bond price movements preceding the forecasting period.47 Alternatively, a representative maturity may be chosen as a general assumption of the interest rate elasticity of traded sovereign exposures. 48 This step ignores the second-order effect of interest rate changes on the future bond price (convexity) in the determination of haircuts.49 Duration is a first-order approximation to the true change in the value of a fixed income security and is only applicable for small changes in yields (less

than 100 bps) which may underestimate the impact of large shocks. A second-order approximation (or “convexity adjustment”) could be added for larger shocks. The convexity is a measure of how the duration of a bond changes as the interest rate changes. Specifically, it is assumed that the interest rate is constant across the life of the bond and that changes in interest rates occur evenly.



for the modified duration

∏=

+ ++ −

+=−

−D

rnc

m rp T trb j k t

b j k tnT t n

b j k tn m

b j k tT t

1(1 ) (1 )

( )(1 )

., [ ],, [ ],

1, [ ], , [ ],

1

1 1 1

(A9.2.6)

Given the zero- coupon bond pricing formula

( )(1 ), [ ], [ ], [ ] [ ],2P exp r LGD PDb j k t j k t j k j k T t= − − − (A9.2.7)

with a cumulative probability of default (PD)

(1 (1 ) )[ ], [ ],PD PDj k T t j k t

T t= − −−−

(A9.2.8)

at the last observable sample date t until maturity date T, and given constant loss given default (LGD) and the unknown country- specific risk- free rate rf j k t[ ],

, we can re- write the ZCE to

Pp

rexp r LGD PDb j k t

ZCE

b j k tT t t f j k j k T tj k t(1 )

ˆ (1 ), [ ],, [ ],

[ ] [ ],1

1

[ ],( )=+

+ ϕ = − −− −

(A9.2.9)

assuming a constant hazard rate in continuous time. Using the sovereign CDS as a measure of country- specific default risk, the equation above can be rewritten as

Pp

rexp r

sT tb j k t

ZCE

b j k tT t t f

CDS

j k t

j k t

(1 )ˆ

10,000( ), [ ],

, [ ],1

1

[ ],

[ ],=+

+ ϕ = − +

−

−

(A9.2.10)

where

1(1 ) 10, 000[ ], [ ][ ],

sT t

ln PD LGDCDS j k T t j kj k t= −

−− ×−

(A9.2.11)

is the cash k- year sovereign CDS spread (in basis points) of country j at time t, which represents idiosyncratic credit risk.Equation A9.2.10 above can now be solved for the time- varying risk- free rate

rT t

lnp

r

s sf

b j k tT t t

CDS CDS

j k t

j k t j k tˆ 1(1 )

min( , )

10,000, [ ],[ ],

1

[ ], [ ],=− +

+ ϕ

−

−

(A9.2.12)

after the market- implied credit risk is smoothed. The default risk is defined as the minimum of the last observable cash CDS spread with a maturity of k years, sCDS j k t[ ],

, and the average cash CDS spread over the last 12 months of the estimation period, sCDS j k t[ ],

.Some flexibility is also introduced in the selection of time t, which might occur before the year- end (a commonly used refer-

ence date for stress test scenarios). Thus, the yield is decreased by ∆ = −+τr r rb j k b j k t b j k t, [ ] , [ ], , [ ],2 2 2 as a market value adjustment to

reflect the pricing effect of shortening the residual maturity to T t− − τ caused by fraction of up to one year τ ∈]0,1[ until time t + τ. So equation A9.2.12 can be generalized to

rT t

lnp

rr

s sf

b j k tT t t b j k

CDS CDS

j k t

j k t j k tˆ 1(1 )

min( , )

10,000, [ ],, [ ][ ],

1

2

[ ], [ ],=− − τ +

+ ϕ

− ∆ −

+τ− −τ +τ+τ

(A9.2.13)

with

PD r r

C r rt b j k t

b j k tcoupon

b j k t b j k t

b j k tcoupon

b j k t b j k t

( )12

( ), [ ],

, [ ], , [ ], , [ ],

, [ ], , [ ], , [ ],21

1 2 1

1 2 1

ϕ =−

+ −

+τ +τ

+τ +τ +τ

+τ +τ +τ

(A9.2.14)

where rb j k t +τ, [ ],1 and rb j k t +τ, [ ],2 are the updated yields reflecting the valuation effect (caused by the mismatch of the desired start-ing point and the timing of the last observable bond price), and Pb j k t +τ, [ ],1

is the updated market price of the coupon bond (without coupon payments). Given τ ≥ 0 , equation A9.2.13 above can then be solved for the time- varying risk- free rate.



Calculation of the credit risk premium under baseline conditions

For the determination of the future bond price under baseline conditions, observable market expectations about changes in country- specific default risk are incorporated. The future price Pb j k i, [ ],2

of each outstanding bond of country j at time t i+ τ + is calculated over a stress test horizon of T t i− − τ − years, where i I∈ . It is derived from using the estimated risk- free rate rf j k t +τ

ˆ[ ],

and applying the ith- period maturity- matched market- implied default risk PDj k i[ ], to the standard zero- coupon pricing formula so that

P exp

r r

maxs s

f fT tb j k i

f f

CDS CDS

CDS CDS

j k t j k t

j k t j k t

j k i j k i

ˆ ˆ

min( , ),

min( , )

10,000( ), [ ],2

[ ], [ ],

[ ], [ ],

[ ], [ ],= − +

+ ∆

− − τ

+τ

(A9.2.15)

where P Pb j k i b j k t, [ ], , [ ],2 2− +τ informs the haircut relative to the current valuation.

The implied default risk [ ],PDj k i for each period of the test horizon is given by the continuous time expression of default

PD exp

maxs s

f fT t

LGDj k i

CDS CDS

CDS CDS

j k

j k t j k t

j k i j k i

min( , ),

min( , )

10,000( ) 1

1[ ],

[ ]

[ ], [ ],

[ ], [ ],= −

− − τ

−

(A9.2.16)

based on the hazard rate PD LGDj k i j k[ ], [ ]× , where the k- year sovereign CDS spread at the ith- period is defined as the greater of (1) the minimum of the last observable cash CDS spread, sCDS j k t[ ],

(with a maturity of k years prior to the starting point of the forecasting period) and the average cash CDS spread, sCDS j k t[ ],

(over the last 12 months of the estimation period); and (2) the minimum of the last observable forward CDS spread, fCDS j k t[ ], (with a maturity of k years prior to the starting point of the forecasting period), and the average forward CDS spread, fCDS j k t[ ],

(over the last 12 months of the estimation period) (see Ap-pendix Box 9.2.1).50,51 The forward CDS spread can be interpreted as the price effect of uncertainty around the expected default risk expressed in the cash CDS spread of the same maturity.52 The term rf j k t

ˆ 0[ ],

∆ > denotes the possibility of incorporating a positive (common) shock to the risk- free rate during all (or selected) periods during the test horizon. This interest rate shock is kept maturity- specific to account for either an upward (parallel) shift or a change in the slope of the interest rate term structure. In the case of the former, it would simplify to rf j t

ˆ 0,

∆ > .For comparative purposes, the same approach is applied to the DCF pricing formula, which is used for the estimation of

market risk parameters in EU system- wide stress tests. Based on equation A9.2.2, we obtain

∏=+ + θ

++ + θ

≥=− −τ

− −τ(1 ) (1 ), [ ], 1, [ ], [ ], , [ ], [ ],

, [ ],11 1

2P

cr

pr

Pb j k i nT t n

b j k t j k tn m

b j k t j k tT t b j k i

(A9.2.17)

where

ˆ

max min , min , ,0

10,000,[ ], , [ ][ ], 1

[ ], [ ], [ ], [ ],r r

f f s sj k t f b j k

CDS CDS CDS CDS

j k t

j k i j k i j k t j k t( )( ) ( )θ = ∆ + ∆ +

−

(A9.2.18)

which comprises the same components as equation A9.2.15 by defining the common interest shock, ˆ[ ],

rf j k t∆ (in addition to the

risk- free rate included in rb j k t, [ ],1), the market value adjustment, r r rb j k b j k t b j k t, [ ] , [ ], , [ ],1 1 1

∆ = −τ+ (if the starting time of the test hori-zon is moved beyond the empirical cut- off date), and the expected increase of sovereign risk implied by the forward CDS spread. In contrast to equation A9.2.15, however, this specification includes only the marginal increase of country- specific risk (as specified in equation A9.2.18) since the uncertainty associated with the expected default risk at time t, min ,

[ ], [ ],( )s sCDS CDSj k t j k t

is already reflected in the observable bond price Pb j k i, [ ],1.53

50 Using the current CDS spread as the lower bound in this specification assumes away valuation gain.51 Because of certain simplifying assumptions, especially regarding the pricing of the forward CDS term structure, the estimates of each country’s haircuts

in outer years are biased downward.52 For instance, the forward matrix for the sovereign CDS spread at the test horizon of {0.5,1,3,5,7,10}i I∈ = years can be obtained from Bloomberg LP via

the command “BDS(“[add ticker of CDS],” “FORWARD_CDS_MATRIX”,”SW_CURVE_DT=20161231”,”startrow=”&2,”endrow=”&2,”cols=10; rows=1”)” on December 31, 2016, where “[add ticker of CDS]” would need to be replaced with “GERMAN CDS USD SR 5Y Corp” to generate the in-formation for senior CDS on German government bonds at a maturity of five years.

53 When individual bond data is not available, the average duration of a given sovereign bond portfolio could be used as an approximation. Then bond valuation changes would be defined by / / (1 ), , , , , , , , , ,1 1 1 1 1

dP P D r rb j t b j t b j tcoupon

b j t b j t− = + × ∆ , where ˆ , ,1rb j t∆ is the change in volume- weighted sovereign yield, and

, ,1Db j t

coupon is the average duration of the bond portfolio.



Appendix Box 9.2.1. Forward Credit Default Swap (CDS) Contracts and the Standard CDS Pricing Formula

This box derives a market model for forward rates on CDS contracts consistent with the general pricing model (such as the forward CDS matrix specified with command <FWCS> in Bloomberg) to derive a forward- looking measure of market- implied default risk as specified in equation A9.2.15.

A CDS represents an agreement between two parties, which is typically designed to transfer to the “credit protection seller” the finan-cial loss the “credit protection buyer” would incur if a designated third party (that is, the reference entity) were to default (that is, causing a so- called credit event).

As a natural starting point for the conventional definition of a CDS, consider a contract valid during the time interval [ , ]T Ta b , where the protection buyer pays the annual rate ( )s tRCDS (in basis points) at times { , , , ..., }1t T T T Ta a i b∈ + or until default time T Ta b[ , ]τ ∈ of the reference entity (based on a predefined default event). The seller of the CDS contract then provides the deterministic protection payment LGD = (1−R) at the default time τ . Formally, the discounted cash flows for the protection seller associated with the basic structure of this so- called “running CDS” (RCDS) with unit notional at time t Ta≤ can be written as the net payoff

1∏ ∑= α τ≥= +t D t T s ti i RCDS Ti a

b

premium leg

RCDS i� �� ( ) ( , ) ( ) { }1

D t T s t D t LGDRCDS T T

accrual leg

T T

protection leg

t t t t( , )( ) ( ) ( , ) 01 { } { }1 1

1 1� ��

+ τ − τ − τ =τ+ <τ< <τ≤− −

(A9.2.1.1)

where iα denotes the fraction of one year between Ti and 1Ti− , T 1τ+ is the first date after τ within the time grid Ti , ( )r tf is the prevailing risk- free rate, and D(t) is the discount factor. The general market model underpins the standard pricing methodology accepted by the market (and used in Bloomberg LP), which applies a ( time- changed) Poisson process for the expected default probability while treating the recov-ery rate on default as constant and exogenous. Equation A9.2.1.1 above can be rewritten to the no- arbitrage pricing condition for the closed- form valuation of a CDS contract over a maturity tenor of n days as

� �� ( ) 6

360( ) ( ) ( )

360( ) ( ) ( )1

11

11

PV tt t

s t S t D tt t

s t D x h x dxi iRCDSi

n

premium leg

i iRCDSi

n

t

t

accrual leg

RCDSi

i∑ ∑∫= − + −−=

−=−

� ��

( ) ( ) ( ) 0exp h x dx LGD D x h x dxt

t

t

t

protection leg

i i∫ ∫( )− − =

(A9.2.1.2)

so that the present value of payoffs to the buyer and seller of credit protection cancel out, where h(t) is the constant “hazard rate” (or de-fault probability), and the survival rate is defined as

( ) ( ) ( ).

01S t exp h x dx exp ht

t

∫( )= − = −

(A9.2.1.3)

Thus, the premium payment of spread ( )s tRCDS to offset the cumulative probability of default (PD) at the last observable sample date until maturity for constant LGD is defined as

s texp h x dx LGD D x h x dx

t tS t D t

t tD x h x dx

ln R PD t tRCDSt

t

t

t

i i i ii

n

t

t

i

n

i i

i

i( )

( ) ( ) ( )

360( ) ( )

360( ) ( )

(1 (1 ) ( ))/ .1 1

11 1

∫ ∫∑∫∑

)(=

−

− + − ≈ − − −− −

== −

(A9.2.1.4)

The first term (“premium leg”) of equation A9.2.1.2 quantifies the quarterly premium amount (in basis points) to be paid to the credit protection seller as the difference in days between payment dates divided by 360 and multiplied by the CDS spread ( )s tRCDS (or “premium coupon”), then multiplied by the probability that the reference entity survives up to the premium payment date and discounted to the present.2

The second term in the formula (“accrual leg”) defines the value of any accrued premium payment on a credit event that occurs be-tween payment periods as the number of days since last coupon date divided by 360, multiplied by the premium amount, multiplied by the conditional probability that a default occurs at time t, and discounted to the present. Any accrued but unpaid premium is paid upon the triggering of a contingent payment after a credit event has occurred.

The final term (“protection leg”) above specifies the value of the payment to the credit protection buyer of LGD, which is the par value (set to unity) minus the recovery value of the reference entity at time t between the effective and scheduled termination date. Such pay-ment after recovery is assumed to occur with the conditional probability of a credit event—that is, the probability that the reference entity survives up to time t multiplied by the hazard rate, discounted to the present.3

1 The input data required are the effective date, maturity date, and premium payment dates; and the risk- free interest rate term structure (derived from Libor/Euribor and swap rates), the survival probability curve obtained from the hazard rate, and the recovery rate for a credit event.

2 Premium payments are calculated on an actual/360 day- count convention.3 In practice, a 30-day delay in the payment of the recovery amount is assumed in the context of this valuation approach. The fair market CDS

spread is calculated with an effective date equal to the trade date plus one business day, and regular quarterly premium payments on March 20, June 20, September 20, and December 20.

(continued)



The general CDS pricing formula above can be adapted to derive the forward rate on the CDS spread. First, equation A9.2.1.1 above is simplified by setting = ττ+1T , which eliminates the accrual leg so that

(A9.2.1.5)

denotes the valuation of the “postponed payoff RCDS” (PRCDS) (Brigo 2004), with

1 1∑= α<τ≤ = + τ≥−

s t D t T LGD D t TPRCDS i T T ii a

bi Tt t i

( ) ( , ) / ( , ) .{ } 1 { }1 (A9.2.1.6)

Since ∏ tPRCDS ( ) is not a real- world payoff (but approximates a general CDS payoff), the CDS price can be computed according to risk- neutral valuation. Under no arbitrage conditions, the discounted payoff of selling the CDS contract can be written as risk- neutral expectation

∏( )Ε tt PRCDSa t�

( ) | , (A9.2.1.7)

which is conditional on the filtration t representing all available information up to time t, where � is the risk- neutral equivalent martingale measure, and default is modeled as an t -stopping time (Brigo and Morini 2005). It is convenient to express prices in credit risk valuation using a subfiltration structure, since a market operator might have information on the probability of default but cannot say exactly when, or even if, default has happened.

Following Jeanblanc and Rutkowski (2000), the flow of all information can be represented except the default itself (“ default- free infor-mation”) in subfiltration t . Thus, it is then possible (and in many cases preferable) to define pricing formulas in terms of the conditional survival probability Pr( | )t tτ > , which can be assumed to be strictly positive in any state of the world, with expectations conditional on the usual default- free filtration t (Brigo and Mercurio 2006) so that4

( , ( ), ) 0PV t s t LGD

PRCDS=

(A9.2.1.8)

where

1

1

∏ ∑

∑

= τ >α Ε

− Ε

τ>τ≥= +

<τ≤= + −

t ts t D t T

LGD D t T

PRCDSt

ti PRCDS t i T ti a

b

premium leg

t i T T ti a

b

protection leg

i

i i

� ��

� ��

�

�

( )1

Pr( | )( ) ( ( , ) | )

( ( , ) | ).

{ }{ }1

{ }1 1

(A9.2.1.9)

If the CDS spread s tPRCDS ( ) is fixed at time t such that the contract in equation A9.2.1.9 above has a value of zero over time steps [ , ]i a b∈ , the following can be written:

1

1

∑∑

=Ε

α Ε

<τ≤= +

τ≥= +

−s tLGD D t T

D t TPRCDS

t i T T ti a

b

i t i T ti a

bi i

i

�

�( )

( ( , ) | )

( ( , ) | )

{ }1

{ }1

1

(A9.2.1.10)

Thus, the real-world CDS represents a ratio of survival probabilities if the right probability measure and information flow (as reflected in the filtration choice) are selected in defining conditional default probabilities.

Given the dependence of one- period CDS spreads on the default probability in the context of real market discrete- tenor CDS spreads, a probability measure associated with discrete tenor interest rates—that is, a forward rate measure—needs to be considered. This would separate the risk- free interest rate from default probabilities. Thus, for the time interval [m,b], the forward rate on the CDS contract under the general market model (see equation A9.2.1.10) is defined as

1 1

1

∑∑ ( )( ) ( )

=Ε − Ε

α Ε

<τ τ≥= +

τ≥= +

−f tLGD D t T H D t T

D t TPRCDS

t i T t t i T ti m

b

j t j T tj m

bi i

j

� �

�( )

( , ) | ( , ) |

( , ) |

{ } { }1

{ }1

1

(A9.2.1.11)

where ( ( , ) ( ; , ) | )1 1� D t T F T T Tt m m m m tΕ − − with the forward measure F(·). In the main text, the forward rate on sovereign CDS spreads

[ ],sCDSj k i

with maturity of k years for i number of years is denoted as

[ ],fCDSj k i

.5

1 1∏ ∑= α − =τ≥= + <τ≤−t D t T s t D t T LGDPRCDS i i PRCDS Ti a

b

premium leg

i T T

protection leg

i t t� �� ( ) ( , ) ( ) ( , ) 0{ }1 { }1

Appendix Box 9.2.1. (continued)

4 For the proof, see Brigo and Mercurio 2006.5 Parties to forward agreements need to have exactly opposite hedging interests that coincide in the timing and amount of the protection bought

and sold against adverse price movements (“double coincidence”). Forward contracts have zero value at the time of inception (that is, they coin-cide with the spot rate). They gain in value as changes to the parameters determining the price of the reference assets increase the future price above the expected price set at inception.



Estimation of the credit risk premium under adverse scenarios

For the specification of valuation haircuts under adverse scenarios, changes in sovereign risk during times of stress are derived from the past dynamics of forward- looking estimates of market- implied assessment of default risk using forward CDS spreads. For this approach, extreme value theory (EVT) is applied as a general statistical concept to model the historical distribution of these spreads to account for large (nonlinear) fluctuations in sovereign risk in the past and the possibility of such tail events affecting the valuation of sovereign bonds over a given test horizon.

More specifically, the historical distribution of spreads is assumed to fall within the domain of attraction of the generalized extreme value (GEV) distribution as a closed form solution to estimating their limiting (or asymptotic tail) behavior—that is, the probability of large positive increases in both the level and volatility of spreads. Based on the parametric fit of the GEV to the historical observations of forward CDS spreads, we can use the density distribution at a high percentile level, such as the conditional tail expectation (CTE), as a stressed country- specific risk component to replace min ,

[ ], [ ],( )f fCDS CDSj k i j k i in the

definition of future bond prices Pb j k i, [ ],1 and Pb j k i, [ ],2

under zero- coupon and DCF pricing, respectively, at time − − τT t as specified in equations A9.2.15 and A9.2.17 above.

This parametric approach is used to model the adverse scenario of valuation haircuts. For each sample country, the indi-vidual asymptotic tail behavior of a historical series of forward CDS spreads is specified through parametrically fitting a se-quence of normalized extremes (maxima or minima) drawn from a sample of independent and identically distributed random variables to a GEV distribution. This enables us to identify the possible limiting laws of asymptotic tail behavior (that is, the likelihood of even larger extremes as the level of statistical confidence approaches certainty). The Fisher- Tippett- Gnedenko theorem (Fisher and Tippett 1928; Gnedenko 1943) defines the attribution of a given distribution of normalized maxima (or minima) to be of an extremal type.

Let the matrix

= f ff CDS CDS

zCDS j k i j k i j k i

, ... ,1[ ], [ ], [ ],

(A9.2.19)

denote a vector- valued independent identically distributed random series of forward CDS spreads fCDS j k i[ ], with a maturity tenor of k years for sample country j at the ith period of the test horizon. y max f ff CDS CDS

zCDS j k i j k i j k i[ ], [ ], [ ],

( ,...,= 1 )) with the cumulative distribution function ( )x and ∈x defines the sample maxima with ascending order statistics f fCDS

zCDSz z

j k i j k i...,1 ,

[ ], [ ],≤ ≤ over an estimation period of z number of observations. The distribution of normalized extremes satisfies

the conditions of GEV if there exists a choice of normalizing constants β > 0[ ],j k iz and α > 0[ ],j k i

z , such that the probability of each ordered z- sequence of normalized sample maxima ( )− α β >f j k i

zj k iz

CDS j k i / 0[ ], [ ],[ ],

converges to the nondegenerate limit distribution ( )

[ ],.G fCDS j k i

as z → ∞ and �[ ],

∈fCDS j k i

,54 so that

lim /

[ ],[ ], [ ],z f j k i

zj k izPr x

CDS j k i→∞−( ) ≤

α β → G fCDS j k i[ ],

( )..

(A9.2.20)

If the normalized extremes only roughly follow GEV, they are considered to fall within the maximum domain of attraction of ( )

[ ],.G fCDS j k i

. In this case, their distribution conforms to one of three distinct types of extremal behavior as limiting distri butions (which are expressed below in their general form without specific notation):55

( )= − − ≥ ξ =0 : ( ) ( ) 0, 00EV G x exp exp x if x

(A9.2.21)

( )= − ∈ µ − σ ξ ∞ ξ >− ξ1: ( ) [ / , [, 011/EV G x exp x if x (A9.2.22)

2 : ( ) ( ) ] , / , ], 0.2

1/( )= − − ∈ − ∞ µ − σ ξ ξ <− ξEV G x exp x if x

(A9.2.23)

If ξ > 0 , GEV falls within the class of Fréchet (EV1) distributions, which feature regularly varying tails, including fat- tailed distributions, such as Stable Paretian distributions. ξ < 0 indicates (negative) Weibull (EV2)-type distributions—that is, distri-butions without a tail but a finite end- point (for example, uniform or beta distributions). In the case of 0ξ → , GEV ap-proaches a Gumbel (EV0) distribution, which encapsulates thin- tailed distributions,56 for which all moments exist.

54 The upper tails of most (conventional) limit distributions (weakly) converge to this parametric specification of asymptotic behavior, irrespective of the original distribution of observed maxima (unlike parametric value at risk models).

55 See Embrechts, Klüppelberg, and Mikosch 1997; Coles 2001; Vandewalle, Beirlant, and Hubert 2004; and Thérond and Ribereau 2012 for additional information on the definition of EVT.

56 For instance, normal, log- normal, gamma, and exponential distributions.



The cumulative distribution functions in the above equations are combined into a unified parametric specification of the general GEV cumulative distribution function,57 which for x fCDS j k i[ ],

is defined as

G x

expx

f

j k i j k i

j k

CDS j k i[ ],( )

( )[ ], [ ],

[=

− +−

1ξ µ

σ ]],

/

[ ]

[ ],

i

j k

j k i

exp expx

− −−

−1 ξ

µ ,,

[ ],

i

j k iσ

if 1+−ξ µj k i j k ix[ ], [ ],( ))

, ,

[ ],

[ ],

σ

ξ

j k i

j k ix

≥

∈ =

0

0if �

(A9.2.24)

with the index for the test horizon dropped from this notation for simplicity. Differencing equation A9.2.24 above as

G xddx

G xf fCDS j k i CDS j k i( ) ( )

[ ], [ ],′ = yields the probability density function

=σ

+ξ − µ

σ

− +

ξ − µσ

− ξ −

+

− ξ −

( )1

1( )

1( )

[ ],

[ ], [ ],

[ ],

1/ 1

[ ], [ ],

[ ],

1/ 1

[ ],

[ ], [ ],

g xx

expx

fj k i

j k i j k i

j k i

j k i j k i

j k iCDS j k i

j k i j k i

(A9.2.25)

where the scale, location, and shape parameters are estimated as µ >ˆ 0[ ],j k i , σ >ˆ 0[ ],j k i , and ξ̂ [ ],j k i, respectively.58 The scale param-

eter represents the annualized volatility of (at least) monthly observations of CDS spreads. The shape parameter is determined by the type of sub- model (EV0, EV1, or EV2). The moments are estimated concurrently by means of the linear combinations of ratios of spacings method, which determines how quickly the probability of extreme observations converges to zero, using the historical spread dynamics over a chosen estimation horizon (Coles 2001; Jobst 2007) (see Appendix 9.4).59 The associated maximum likelihood estimator is evaluated numerically by using an iteration procedure (for example, over a rolling window of a constant number of observations with periodic updating) to maximize the likelihood ∏ θ= ( | )1 [ ],

g xfiz

CDS j k i over all three

parameters θ µ σ ξ= ( ˆ , ˆ , ˆ )[ ], [ ], [ ],j k i j k i j k i simultaneously.60 Given the expectation

∫ σ+

ξ − µσ

− +

ξ − µσ

− ξ − − ξ∞ x x

expx

dxj k i

j k i j k i

j k i

j k i j k i

j k i

j k i j k i

ˆ1

ˆ ( ˆ )ˆ

1ˆ ( ˆ )

ˆ[ ],

[ ], [ ],

[ ],

1/ ˆ 1

[ ], [ ],

[ ],

1/ ˆ

0

[ ], [ ],

= µ +σ− ξ

− +

ξ − µσ

− ξx

j k ij k i

j k i

j k i j k i

j k i

j k i

ˆˆ

1 ˆ 1ˆ ( ˆ )

ˆ[ ],[ ],

[ ],

[ ], [ ],

[ ],

1/ ˆ[ ],

(A9.2.26)

based on the cumulative distribution function in equation A9.2.24 above, we obtain the CTE (or conditional VaR) as probability- weighted residual density beyond a prespecified statistical confidence level (“severity threshold”) over the certain estimation period. The corresponding density distribution at a certain statistical confidence level a can be derived as

CTE x x G a VaRa f fCDS j k i CDS j k i, [ ], [ ],

| ( )= ≥ =−E 1aa fCDS j k i, [ ],( ) (A9.2.27)

57 Standard pricing models of CDS spreads assume that the essential input variables (that is, the likelihood of default and the recovery rate up to one year) provide sufficient statistical support to inform a multiperiod estimate of expected loss as a monotonically increasing density function of continuously distributed default risk.

58 The upper tails of most conventional limit distributions weakly converge to this parametric specification of asymptotic behavior, irrespective of the origi-nal distribution of observed maxima (unlike parametric VaR models). The higher the absolute value of the shape parameter, the larger the weight of the tail and the slower the speed at which the tail approaches its limit. The shape parameter also indicates the number of moments of the distribution—for example, if ξ = 1/2, the first moment (mean) and the second moment (variance) exist, but higher moments have an infinite value. The moments of order

≥ ξn 1/ are unbounded—that is, ξ1/ indicates the highest bounded moment for the distribution. This is of practical importance since many results for asset pricing in finance rely on the existence of several moments.

59 A rough estimation for each sample country is provided in the data template for this method (available in MS Excel® file “Data_Input.xlsx” (available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit), which allows the user to download the relevant CDS data from Bloom-berg L.P. and calculate the respective point estimates of changes in CDS spreads according to the quantile function G afCDS j k i

( )1[ ],

− in equation (A9.2.28). A uniform shape parameter of 1ξ ≤ − = 0.33 is applied in this calibration approach to generate the results shown in Appendix 9.5.

60 The maximum likelihood estimator fails for 1ξ ≤ − since the likelihood function does not have a global maximum in this case. However, a local maxi-mum close to the initial value can be attained.




with quantile function

( )( )= µ +

σξ

− −− −ξ( ) ˆ

ˆˆ ( ) 11

[ ],[ ],

[ ],

ˆ

[ ],

[ ],G a ln af j k ij k i

j k iCDS j k i

j k i

(A9.2.28)

and

VaR sup G Pr x Ga f f fCDS j k i CDS j k i CD, [ ], [ ],( )= ⋅ >−1

SS j k ia

[ ],( )− ⋅( ) ≥( )1 ,

(A9.2.29)

where

G a G a inf x Gf f fCDS j k i CDS j k i CDS j[ ], [ ],( ) ( )− ←= ≡1

[[ ],

[ ],

[ ],k i

xaj k i

z

j k iz

−−

≥

1

αβ . (A9.2.30)

Equation A9.2.27 is specified by the general definition of CTE (Artzner and others 1999) as

CTEx dx

a f

VaR

CDS j k i

a fCDS j k i

, [ ],

, [ ],

( ( ))=

−∞

∫ 1

11

11− ( ) =

−F VaR aVaR da

a f

a f

CDS j k i

CDS j k i

,

,

[ ],

[ ],aa

1

∫ ,

(A9.2.31)

where VaRa f x F x aCDS j k i, inf ( | ( ))

[ ],≡ ≥ is the quantile of order 0 < a < 1 (say, a = 0.95).

Thus, we can refine the specification of the future price of each outstanding bond of country j (with a common shock to the interest rate term structure at period i) under both pricing approaches— equations A9.2.15 and A9.2.17 above— as

ˆ( ) ˆ ˆ10,000

( ), [ ],

,

2 [ ], [ ],

[ ],P a exp r rCTE

T tb j k i f f

a f

j k t j k t

CDS j k i= − + ∆ +

− − τ

τ+

(A9.2.32)

and

ˆ( )(1 ˆ( ) ) (1 ˆ )

, [ ], 1, [ ], [ ],

/, [ ], [ ],

1

1 1

P ac

r a

p

rb j k i n

T t n

b j k t j k in m

b j k t j k iT t

= ∏+ + θ

++ + θ

ττ=

− −− −

(A9.2.33)

where

( )( )θ = ∆ + ∆ +

−ˆ( ) ˆ

max min , ,0

10,000[ ], [ ], , [ ]

,

1

[ ], [ ], [ ],a r r

CTE s sj k i j k t b j k

a f CDS CDSCDS j k i j k t j k t .

(A9.2.34)

The valuation haircuts are then derived from bond price changes in response to expected changes in idiosyncratic (default) risk and common interest rate shocks. For different scenarios (affecting the severity of haircuts), we distinguish between (1) current market expectations (for the baseline scenario) based on the prevailing level of forward sovereign CDS spreads, fCDS j k i[ ], ; and (2) the density forecasts of expected default risk (for the adverse scenarios), G afCDS j k i

( )1[ ],

− , based on the historical dynamics of forward sovereign CDS spreads, whose empirical distribution has an asymptotic tail behavior consistent with GEV as defined in equation A9.2.28 above: 61

• For the baseline scenario, the spreads observed at the start of the stress test horizon imply the expected change of default risk affecting the future bond price over i- periods in the future based on the two different pricing formulas in equations A9.2.15 and A9.2.17 above.

• For the adverse scenarios, however, haircuts should reflect the volatility of market expectations of default risk; thus, point estimates are derived at high levels of statistical confidence to project the model- based impact of higher spreads during times of stress on bond prices. For instance, density forecasts can be chosen at the 75th percentile (for a mild adverse scenario [“adverse 1”]) and 90th percentile (for a severe adverse scenario [“adverse 2”]) of the quantile function

61 Confidence intervals were obtained, on which basis an adverse scenario can be constructed (for example, one standard deviation worse than the baseline, or, preferably, at a very level of statistical significance).



(see equation A9.2.28) as country- specific shocks.62 The percentile choice should be consistent with the probabilistic severity of the overall macroeconomic scenario of the stress test; however, the statistical significance of country- specific spread shocks cannot be directly compared to that of changes in the real GDP growth rate (or other, low- frequency macroeconomic variables), which often inform the severity of the macroeconomic scenarios. Given the shorter estima-tion horizon and higher frequency of CDS data, the percentile level of the density forecast tends to be lower than the one implied by the projected deterioration of macroeconomic conditions over the stress test horizon.

Thus, for each year over the test horizon of i n∈ years, there is a vector of three bond prices

P { }= P P a P ab j k i b j k i b j k i b j k iadverse adverse

; ˆ( ) ; ˆ( ), [ ], , [ ], , [ ], , [ ],1 1 1 1 1 2 (A9.2.35)

and

P { }= P P a P ab j k i b j k i b j k i b j k iadverse adverse

; ˆ( ) ; ˆ( ), [ ], , [ ], , [ ], , [ ],2 2 2 1 2 2 (A9.2.36)

for each pricing method, based on current market expectations and two different density forecasts of default risk at statistical confidence level a {0.75;0.90}∈ .

The corresponding haircuts are calculated for each bond from changes in bond prices in each year i over the test horizon, relative to the base year t, using the following specification for the baseline scenario as

∆P

P

Pb j k ib j k i

b j k tZCE1

1

1

1, [ ],, [ ],

, [ ],

= −

××100

(A9.2.37)

and

∆P

P

Pb j k ib j k i

b j k t2

2

2

1 10, [ ],, [ ],

, [ ],

= −

× 00

(A9.2.38)

where Pb j k i, [ ],1 and Pb j k i, [ ],2

are the bond prices under each pricing method, respectively.63 The general haircut h for country j is derived as an issuance size- weighted average of individual projected haircuts applied to a q- number of bonds outstanding,64 so that

, 0, [ ],

, [ ],

, [ ],1

, [ ],1

,

,1

1

2

1

2

hh

maxP

P

Amt

Amtb j k i

b j k i

b j k ib

q

b j k ib

qb j

b jb

q

∑∑ ∑

=∆∆

×

=

= =

,

(A9.2.39)

where Pb j k i, [ ],1∆ and Pb j k i, [ ],2

∆ are the haircuts under each pricing method over test period i, and Amtb j, is the outstanding amount of bond b issued by country j. As a final step, these haircuts would then be applied to the amount of sovereign bond exposures to countries j J∈ held in both the banking and trading books at time t. The sovereign bond losses or changes in

valuation in each year t over the test horizon are calculated as hh exposureb j k i

b j k ijJ

i j, [ ],

, [ ],,

1

2

∑

× , based on a firm’s total exposure to country j.

62 The calibration of the country- specific spread shock is based on annualized volatility of (at least monthly) observations. This means that point estimates at a chosen percentile level imply a much higher degree of statistical confidence (and, thus, are more extreme) than growth shocks at the same percentile level (derived from historical distribution of annual growth rates). In practical terms, the country- specific sovereign risk shock at the 75th percentile is likely to be consistent with a decline of real GDP growth of at least twice its standard deviation (or 98th percentile) over a long- term estimation period (which is the statistical confidence underpinning most adverse scenarios in system- wide stress tests (Jobst, Ong, and Schmieder 2013).

63 The haircut estimation is not fully accurate because the bond portfolio is assumed to be constant (that is, without replacement of maturing bonds with newly issued securities). The assumption overstates the actual haircut, unlike in cases when the sample of bonds changes, and the remaining maturity is kept constant.

64 Haircuts are set to zero when bond prices rise (for example, for safe- haven sovereigns).



1. Germany

Realized distributionover projection

horizon

Empirical distributionover estimation horizon(Fitted GEV distribution)

0

250

50

100

150

200

Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)

2. France


horizon



horizon



horizon


Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)

0

450

50100150200250300350400

3. Italy

100200300400500600700800

0

900

Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)4. United Kingdom

Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)

0

400

50100150200250300350

Appendix Figure 9.2.2 Selected Sample Countries: Sovereign Credit Default Swap, Five-Year Maturity Term—Projected and Realized Spreads over Projection Horizon (2011–15)

(In basis points)



5. United States 6. Brazil

0

450

50100150200250300350400

7. Japan

100200300400500600700800

0

9008. Mexico

100200300400500600700800

0

900

0

200

20406080

100120140160180


horizon



horizon


Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)

Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)


horizon


Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)Realized distribution

over projectionhorizon


Spot

* (G

EV-fi

tted)

1YR

FW**

(GEV

-fitte

d)

2YR

FW (G

EV-fi

tted)

3YR

FW (G

EV-fi

tted)

4YR

FW (G

EV-fi

tted)

5YR

FW (G

EV-fi

tted)

Actu

al S

pot (

afte

r 5 y

rs.)

Actu

al S

pot (

afte

r 4 y

rs.)

Actu

al S

pot (

afte

r 3 y

rs.)

Actu

al S

pot (

afte

r 2 y

rs.)

Actu

al S

pot (

afte

r 1 y

r.)

Sources: Bloomberg LP; and authors’ estimates.Note: FW = forward; GEV = generalized extreme value; yrs. = years.* Historical density estimates based on GEV distribution fitted to observed sovereign credit default swap (CDS) spot spreads (with shape parameter = 0.33) according to Appendix 9.1 (with shape parameter = 0.33) over the estimation horizon from March 2009 to December 2010.** Historical density estimates based on GEV distribution fitted to observed forward sovereign CDS spreads (with starting times between one and five years) according to Appendix 9.2 (with shape parameter = 0.33) over the estimation horizon from March 2009 to December 2010. Boxplots include the mean (yellow dot), the 25th and 75th percentiles (gray box, with the change of shade indicating the median), and the 10th and 90th percentiles (whiskers). The blue/green dots indicate the projected CDS spreads under the adverse scenarios (at the 75th and 90th historical density) while the gray dots show the actual CDS spreads at the end of each year during the projection horizon. The red line indicates the highest observed CDS spread observed during the estimation horizon.

Appendix Figure 9.2.2 (continued )


Appendix 9.3.Moments of the GEV Distribution and

Estimation of the Shape Parameter Using the Linear Combination of Ratios

of Spacings Method

Since all raw moments of G (.) are defined contingent on the tail shape, the natural estimator of ξ̂ is derived by means of the Linear Combination of Ratios of Spacings (LRS) method using the linear combination

∑ξ =

−

−

=ˆ

4( ˆ )

( )

1

1

/4n ln vln c

ii

n

(A9.3.1)

for n observations, where vx x c

x c xin a n na n

na n na n

ˆ (1 ): :

: :

=−−

− and cln a

ln a(1 )

( )=

− for quantile a = i/n. Since x G ana n ( ):

1= − , the approximation

≈ − −−

=− −

− −− +ξˆ (1 ) ( )

( ) ( )

1 1

1 11 ˆv

G a G aG a G a

ci

c

c holds. The simple statistics are defined as

mean :

( ),µ σ

ξξ ξ

µ σγ ξ

ξ

+ − ≠ <

+ =

∞ ≥

gif

if

if

1 10 1

0

1

(A9.3.2)

variance

g gif

if:

( ),σ

ξξ ξ

σ π ξ

2 2 12

2

22

012

60

− ≠ <

= ,

∞ ≥

if ξ 1

2

(A9.3.3)

skewness

g g g gg g

if

:

( ) /3 1 2 1

3

2 12 3 2

3 20

− −−

< ξ <<

− − −−

<

=

13

3 20

012

3 1 2 13

2 12 3 2

g g g gg g

if

if

( ) /ξ

ξ66 3

13

3

ςπ

ξ

( )

if ≥∞

, and

(A9.3.4)

Kurtosis

g g g g g g gg g

:

( )4 1 3 2

22 1

214

2 12

4 3 12 6− − − −− 22

014

125

0

14

if

if

if

ξ ξ

ξ

ξ

≠ <

=

∞ ≥

,

,

(A9.3.5)

with = Γ − ξ(1 )g pp for p [1,...,4]∈ , Euler’s constant γ (Sondow 1998), Riemann zeta function (.)ς (Borwein, Bradley, and Crandall 2000), and gamma probability density function (.)Γ .


Appendix 9.4. Credit- Risk- Based Approach

to Estimating Expected Losses of HtM Sovereign Exposures in the

Banking Book

Since held- to- maturity securities in bank books are recorded at historical cost, changes in their market valuation do not impact net income; however, their expected loss should be covered by provisions based on estimated credit risk parameters, the prob-ability of default (PD), and the loss given default (LGD).

The cross- sectional sensitivity of rating- implied PDs to macroeconomic conditions helps determine changes in sovereign default risk under stress (and the associated coverage rate of provisions). The prevailing credit risk assessment of sovereign issu-ers by one or more of the leading credit rating agencies (for example, Moody’s, S&P, or Fitch) can be used to determine the corresponding PD. Mapping tables help convert ratings into PDs over different test horizons. The implied PD obtained this way represents the starting point of sovereign default risk (PD )0 . For instance, based on a panel regression analysis, the elastic-ity can be estimated as

log (PD ) / ,tit drgdptγ = ∆ ∆ (A9.4.1)

where (1) logit PD( )t∆ is the forecasted change in default risk logit PD logit PD( ) ( )t 0− , with logistic transform (logit); and (2) drgdpt is the year- over- year growth rate of real GDP. This expression can be rearranged as:

=−

× γ∆ +−

× γ∆

−

−

−

−

PDPD

PDexp drgdp

PD

PDexp drgdpt

t

tt

t

tt1

( )/ 11

( )1

1

1

1

.

(A9.4.2)

Hence, based on the above elasticity and the changes in real GDP growth under each scenario, the implied sovereign PD can be calculated. If no historical evidence of technical default is available, the recovery rate—that is, 1-LGD—can be extracted from the World Bank’s Doing Business Report (World Bank 2016)65 and used as a proxy for the LGD.66

65 “Doing Business” measures regulations affecting 11 areas of the life of a business. Ten of these areas are included in the 2017 ranking on the ease of doing business: starting a business, dealing with construction permits, getting electricity, registering property, getting credit, protecting minority investors, paying taxes, trading across borders, enforcing contracts, and resolving insolvency.

66 For a cross- country sample of 117 countries, this approach was applied to a total of 2,120 observations. Panel fixed effects were used for the estimation of γ . This elasticity γ was estimated to be −0.09.


Appendix 9.5.Detailed Estimation Results:

Valuation Haircuts




Sovereign Valuation Haircuts—Common and Country-Specific Interest Rate Shocks (Zero-Coupon Pricing/Discounted Cash-Flow Method)

Sovereign Debt Valuation HaircutBaseline Scenario

(Current expectations based on end-year forward prices)

Adverse Scenario (Forecast based on historical density function)

75th percentile 90th percentile

2011 2012 2013 2014 2015 2011 2012 2013 2014 2015 2011 2012 2013 2014 2015IMF-FSAP Approach—Zero-Coupon Pricing Method with Forward CDS Spreads

Europe Austria 4.3 4.5 4.3 4.0 3.9 3.5 3.9 4.0 3.9 3.6 4.5 5.0 5.1 4.9 4.7 Belgium 9.5 9.0 8.2 7.8 7.3 6.3 6.7 6.5 6.1 5.8 10.0 10.6 10.1 9.5 9.3 Finland 3.2 3.3 3.4 3.4 3.4 2.8 3.0 3.1 3.1 3.1 3.0 3.4 3.5 3.5 3.5 France 5.8 5.9 5.6 5.6 5.2 4.4 4.9 4.9 4.7 4.5 6.1 6.7 6.7 6.6 6.4 Germany 4.4 4.7 4.8 4.7 4.7 3.2 3.5 3.7 3.7 3.7 3.8 4.2 4.5 4.5 4.4 Netherlands 4.0 4.3 4.4 4.4 4.4 3.1 3.5 3.5 3.5 3.5 3.8 4.2 4.2 4.2 4.3

Greece 5.9 8.5 12.2 12.2 20.7 13.9 8.4 7.2 9.1 10.1 28.7 21.7 20.4 22.9 24.1 Italy 7.7 7.2 6.8 6.6 6.5 6.2 6.6 6.4 6.1 5.8 10.1 10.7 10.3 9.9 9.4 Ireland 14.3 10.9 9.7 10.0 10.0 13.3 12.1 10.9 10.7 10.6 24.4 22.5 20.7 20.4 20.2 Portugal 12.5 9.9 8.1 6.8 5.7 10.6 9.8 8.4 7.5 7.5 19.4 18.1 15.9 14.7 14.5 Spain 11.1 10.5 9.9 9.2 8.9 8.0 7.6 7.3 7.2 7.0 13.6 12.9 12.6 12.6 12.2

Czech Republic 3.2 3.4 3.5 3.5 3.7 3.6 4.0 4.3 4.4 4.4 4.5 4.9 5.3 5.4 5.5 Denmark 3.6 3.8 3.9 3.9 3.9 3.0 3.3 3.4 3.4 3.4 3.4 3.6 3.8 3.9 4.1 Poland 4.7 5.2 5.5 5.7 5.8 4.8 5.3 5.6 6.0 6.0 6.5 7.0 7.4 7.9 8.1 Sweden 3.4 3.6 3.6 3.6 3.7 3.2 3.5 3.6 3.6 3.6 3.8 4.1 4.3 4.3 4.2 United Kingdom 3.4 3.6 3.6 3.5 3.4 3.6 4.1 4.2 4.0 3.6 4.6 5.1 5.1 4.8 4.2Other Countries Brazil 3.7 4.5 5.1 5.6 5.8 4.4 5.3 5.9 6.3 6.4 5.6 6.5 7.1 7.4 7.4 Japan 4.6 5.4 6.0 6.2 6.3 4.6 5.6 6.4 6.8 7.0 5.9 7.2 8.2 8.8 9.2 Mexico 3.6 4.4 4.8 5.0 5.2 4.3 5.0 5.6 5.8 5.7 5.5 6.2 6.7 6.8 6.7 United States 3.4 3.6 3.7 3.8 3.7 3.1 3.5 3.6 3.7 3.7 3.5 4.0 4.2 4.3 4.5

Adapted EBA Approach—Discounted Cash Flow Method with Forward CDS SpreadsEurope Austria 3.9 4.1 4.0 3.7 3.5 3.6 3.7 3.5 3.3 3.6 4.5 4.6 4.5 4.3 4.7 Belgium 8.5 8.0 7.4 7.0 6.6 6.1 5.8 5.4 5.2 5.8 9.5 9.1 8.5 8.3 9.3 Finland 2.9 3.0 3.1 3.1 3.1 2.7 2.8 2.8 2.8 3.1 3.1 3.2 3.2 3.2 3.5 France 5.4 5.4 5.1 5.2 4.8 4.5 4.5 4.3 4.1 4.5 6.2 6.2 6.0 5.8 6.4 Germany 4.0 4.3 4.3 4.3 4.2 3.2 3.4 3.4 3.4 3.7 3.8 4.1 4.1 4.0 4.4 Netherlands 3.7 3.9 4.0 4.0 4.1 3.2 3.2 3.2 3.2 3.5 3.8 3.9 3.9 3.9 4.3


Czech Republic 2.8 3.0 3.1 3.1 3.2 3.5 3.8 3.8 3.9 4.4 4.4 4.6 4.7 4.8 5.5 Denmark 3.3 3.4 3.5 3.5 3.5 2.9 3.0 3.1 3.1 3.4 3.3 3.4 3.5 3.7 4.1 Poland 4.0 4.4 4.6 4.8 4.9 4.5 4.8 5.0 5.1 6.0 5.9 6.2 6.7 6.8 8.1 Sweden 3.0 3.2 3.2 3.3 3.3 3.1 3.2 3.2 3.2 3.6 3.7 3.8 3.8 3.7 4.2 United Kingdom 3.1 3.3 3.3 3.2 3.1 3.7 3.8 3.6 3.3 3.6 4.6 4.6 4.3 3.8 4.2Other Countries

Brazil 2.8 3.4 3.9 4.2 4.3 4.0 4.5 4.7 4.8 6.4 4.9 5.3 5.6 5.5 7.4 Japan 4.4 5.2 5.7 5.9 6.0 5.4 6.1 6.5 6.7 7.0 6.9 7.8 8.5 8.8 9.2 Mexico 3.0 3.6 3.9 4.1 4.2 4.1 4.6 4.7 4.7 5.7 5.0 5.5 5.5 5.5 6.7 United States 3.1 3.3 3.4 3.5 3.4 3.2 3.3 3.4 3.4 3.7 3.7 3.9 4.0 4.1 4.5Sources: Bloomberg LP; and authors’ calculations. Note: The valuation haircuts show the expected cumulative weighted-average price decline of selected benchmark government bonds over a five-year test horizon (2011–2015) relative to the observed market price on December 31, 2010, based on a common interest rate shock of 50 basis points and a proportionate country-specific credit spread shock according to the zero-coupon pricing formula (“IMF FSAP”) and the discounted cash flow pricing formula used in the EU system-wide stress testing exercises (“EBA Approach”) as specified in Appendix 9.2. The country-specific credit spread shock is derived from the historical changes of forward contracts on five-year sovereign credit default swaps (CDS) until end-2010 (that is, cut-off date). The baseline is based on “current expectations” using the larger of the (1) last observed spot (forward) spread and (2) the average spot (forward) spread over the 12 months prior to the cut-off date. The two adverse scenarios reflect the density distribution (based on GEV-fitted asymptotic tail behavior) of historical spread dynamics at the 75th and 90th percentiles, respectively. The haircuts for European countries under the “IMF Approach” in 2011 [black box] correspond with the values for the baseline and adverse scenarios in Table 9.5. The heavy discounting of bonds issued by vulnerable euro area economies during 2011 implies little (if any) additional haircuts based on market prices beyond the initial test period. CDS = credit default swap; EBA = European Banking Authority; FSAP = Financial Sector Assessment Program; GEV = generalized extreme value.




Sovereign Valuation Haircuts—Country-Specific Interest Rate Shock Only (Zero-Coupon Pricing/ Discounted Cash-Flow Method)

Sovereign Debt Valuation HaircutBaseline Scenario

(Current expectations based on end-year forward prices)

Adverse Scenario (Forecast based on historical density function)

75th percentile 90th percentile2011 2012 2013 2014 2015 2011 2012 2013 2014 2015 2011 2012 2013 2014 2015

IMF-FSAP Approach—Zero-Coupon Pricing Method with Forward CDS SpreadsEurope Austria 2.1 2.2 2.1 1.8 1.6 1.3 1.7 1.8 1.6 1.4 2.2 2.7 2.9 2.7 2.5 Belgium 7.2 6.6 5.9 5.4 5.0 3.9 4.4 4.1 3.7 3.4 7.8 8.3 7.8 7.2 7.0 Finland 0.9 1.1 1.2 1.2 1.2 0.5 0.8 0.9 0.9 0.9 0.8 1.2 1.3 1.3 1.3 France 3.5 3.5 3.2 3.2 2.8 2.0 2.4 2.5 2.3 2.1 3.7 4.4 4.4 4.2 4.0 Germany 2.1 2.4 2.4 2.3 2.3 0.8 1.2 1.4 1.4 1.3 1.4 1.9 2.1 2.1 2.0 Netherlands 1.8 2.1 2.2 2.2 2.2 0.9 1.2 1.3 1.2 1.2 1.5 2.0 2.0 2.0 2.0



Adapted EBA Approach—Discounted Cash Flow Method with Forward CDS Spreads

Europe Austria 1.9 2.1 1.9 1.7 1.5 1.2 1.5 1.6 1.5 1.3 2.1 2.5 2.6 2.5 2.3 Belgium 6.4 6.0 5.3 4.9 4.5 3.5 3.9 3.7 3.3 3.1 7.0 7.5 7.0 6.5 6.3 Finland 0.8 1.0 1.1 1.1 1.1 0.5 0.7 0.8 0.8 0.8 0.7 1.1 1.1 1.2 1.2 France 3.2 3.2 2.9 3.0 2.6 1.8 2.2 2.3 2.1 1.9 3.4 4.0 4.0 3.8 3.7 Germany 1.9 2.2 2.2 2.1 2.1 0.7 1.1 1.2 1.3 1.2 1.3 1.7 1.9 1.9 1.8 Netherlands 1.7 1.9 2.0 2.0 2.0 0.8 1.1 1.2 1.1 1.1 1.4 1.8 1.9 1.8 1.9



Sources: Bloomberg LP; and authors’ calculations. Note: The valuation haircuts show the expected cumulative weighted-average price decline of selected benchmark government bonds over a five-year test horizon (2011–2015) relative to the observed market price on December 31, 2010, based on a proportionate country-specific credit spread shock according to the zero-coupon pricing formula (“IMF FSAP”) and the discounted cash flow pricing formula used in the EU system-wide stress testing exercises (“EBA Approach”) as specified in Appendix 9.2. The country-specific credit spread shock is derived from the historical changes of forward contracts on five-year sovereign credit default swaps (CDS) until end-2010 (that is, cut-off date). The baseline is based on “current expecta-tions” using the larger of the (1) last observed spot (forward) spread, and (2) the average spot (forward) spread over the 12 months prior to the cut-off date. The two adverse scenarios reflect the density distribution (based on GEV-fitted asymptotic tail behavior) of historical spread dynamics at the 75th and 90th percentiles, respectively. The haircuts for European countries under the ‘IMF-FSAP Approach” in 2011 [black box] correspond with the values for the baseline and adverse scenarios in Table 9.5. The heavy discounting of bonds issued by vulnerable euro area economies during 2011 implies little (if any) additional haircuts based on market prices beyond the initial test period. CDS = credit default swap; EBA = European Banking Authority; FSAP = Financial Sector Assessment Program; GEV = generalized extreme value.


Appendix 9.6.Causes for High Sovereign Exposures

in the Banking Sector

Various factors could encourage banks to hold sovereign exposures that strengthen sovereign- bank linkages (Dell’Ariccia and others 2018). The factors include (1) the regulatory incentive hypothesis, (2) the risk- taking hypothesis, and (3) the deficit ab-sorption hypothesis.

• Regulatory incentive hypothesis: Current banking regulations favor investment in “home sovereigns,” which intensifies bank- sovereign linkages within countries. Sovereign exposures are treated as safe assets, which encourages banks to hold more sovereign securities. Sovereign exposures often receive low risk weights (zero percent for local currency- denominated debt),67 which results in low capital coverage for unexpected losses. Sovereign exposures are also ex-empted from large exposure limits.68 Overall, the accounting- based valuation of banks is not fully sensitive to changes in market prices, because they reflect a diverse mix of market and book values of different types of bank sovereign ex-posures. This approach aims to avoid excessive balance sheet volatility and procyclicality of regulatory capital ratios due to full mark- to- market valuation.69

• Risk- taking (or “carry trade”) hypothesis: Banks may hold an excessive amount of “riskier” sovereign securities with higher yields to enhance profits. Doing so becomes more attractive when returns from loans and alternative assets are low,70 and when short- term rates (the cost of carrying) are low.71 Moreover, banks may take an excessive carry trade risk when their capitalization is already low, and risk- shifting incentives encouraging “gambling for resurrection” (due to limited liabil-ity) are strong (Ari 2017). The expectation of bailout or forbearance in the event of a sovereign default could also encour-age excessive carry trades.

• Deficit absorption (or “financial repression”) hypothesis72: Banks may also hold sovereign debt due to cyclical reasons (defi-cit absorption) and public policies. In downturns, banks often act as ready buyers of sovereign debt as lending opportu-nities and private sector asset returns diminish. During times of stress, banks’ capacity to take the role as “contrarian investors” can contribute to realigning market prices of sovereign exposures with fundamentals (especially if market disruptions are excessive or arbitrage opportunities are expensive or not available). From a structural perspective, classic financial repression refers to government policies that pressure banks to hold more government debt to secure fiscal deficit financing. Such measures are often observed in some EMDEs with fiscal, institutional, and governance chal-lenges.73 Altavilla, Pagano, and Simonelli (2017) also find that public, bailed- out, and poorly capitalized European banks responded to sovereign stress by purchasing domestic public debt more than other banks because of moral suasion.74

67 Usually, sovereign exposures that are either denominated in foreign currency or issued by a foreign entity that does not have the highest rating (“AAA”) receive non- zero risk weights. For internal- ratings- based banks, the risk weights may not be zero depending on their internal ratings of sovereigns. How-ever, the internal- ratings- based approach is more prevalent in advanced economies where internal ratings tend to generate very low capital charges.

68 Furthermore, in the euro area, the preferential treatment extends to all euro- denominated sovereign securities issued by the EU Member States (compli-ant with the Basel framework).

69 Moreover, the liquidity coverage ratio under the Basel framework requires banks to hold sufficient high- quality liquid assets against potential cash- flow shortfalls. Sovereign securities are considered safer and more liquid assets than private sector assets. See Grandia and others 2019 for a current assessment of the availability of high- quality liquid assets in the euro area.

70 The European Systemic Risk Board (2015) found evidence that some euro area banks in stressed countries increased their sovereign exposures when do-mestic macroeconomic conditions deteriorated.

71 This applies to banks that have a stable and low- cost deposit base or access to cheap central bank liquidity during times of monetary accommodation.72 De Marco and Macchiavelli (2016) discuss how the political economy has influenced the scope of government debt exposures in Europe.73 In some economies (for example, India), regulations require banks to hold a minimum amount of sovereign securities, which creates captive demand for

public debt. Directed lending— often to state- owned enterprises— is another instrument. The government may also exercise moral suasion to amplify fi-nancial repression.

74 More specifically, their empirical analysis of determinants of banks’ sovereign exposures between 2007 and 2015 revealed that public banks’ purchases of sovereign debt significantly amplified the impact of sovereign stress on their domestic lending (and lending by their foreign subsidiaries in nonstressed countries).



REFERENCESAcharya, Viral. 2018. “Understanding and Managing Interest

Rate Risk at Banks.” Speech at the Fixed Income Money Mar-kets and Derivatives Association (FIIMDA) Annual Dinner, January 15, Mumbai.

Altavilla, Carlo, Marco Pagano, and Saverio Simonelli. 2017. “Bank Exposures and Sovereign Stress Transmission.” Review of Finance 21 (6): 2103–39.

Altman, Edward, Andrea Resti, and Andrea Sironi. 2004. “Default Recovery Rates in Credit Risk Modelling: A Review of the Lit-erature and Empirical Evidence.” Economic Notes 3 (2): 183–208.

Ams, Julianne, Reza Baqir, Anna Gelpern, and Christoph Trebe-sch. 2018. “Sovereign Default.” In Sovereign Debt: A Guide for Economists and Practitioners, edited by S. Ali Abbas, Alex Pien-kowski, and Kenneth Rogoff. Chapter 7. Oxford: Oxford Uni-versity Press.

Ari, Anil. 2017. “Sovereign Risk and Bank Risk- Taking.” IMF Working Paper 17/280, International Monetary Fund, Washington, DC.

Artzner, Philippe, Freddy Delbaen, Jean- Marc Eber, and David Heath. 1999. “Coherent Measures of Risk.” Mathematical Fi-nance 9 (3): 203–28.

BCBS (Basel Committee on Banking Supervision). 2016. “Stan-dards: Minimum Capital Requirements for Market Risk.” Jan-uary, Bank for International Settlements, Basel.

———. 2017a. “The Regulatory Treatment of Sovereign Expo-sures.” December. Bank for International Settlements, Basel.

———. 2017b. “Finalizing Post- Crisis Reforms.” December. Bank for International Settlements, Basel.

———. 2017c. “Standards: Regulatory Treatment of Accounting Provisions— Interim Approach and Transitional Arrange-ments.” March. Bank for International Settlements, Basel.

———. 2018a. “Basel III Monitoring Report.” October. Bank for International Settlements, Basel.

———. 2018b. “Instructions for Basel III Monitoring.” February. Bank for International Settlements, Basel.

Black, Fischer, and Myron Scholes. 1973. “The Pricing of Op-tions and Corporate Liabilities.” Journal of Political Economy 81 (3): 637–54.

Boone, Laurence, and Silvia Ardagna. 2011. “Economics Eurozone— Euro Rendezvous: Big Bang?” July 21. Economic Analysis, Bank of America- Merrill Lynch, London.

Borwein, Jonathan M., David M. Bradley, and Richard E. Cran-dall. 2000. “Computational Strategies for the Riemann Zeta Function.” Journal of Computational and Applied Mathematics 121 (11): 247–96.

Brigo, Damiano. 2004. “Constant Maturity Credit Default Swap Pricing with Market Models.” December. Working Paper, De-partment of Mathematics, King’s College, London.

Brigo, Damiano, and Fabio Mercurio. 2006. Interest Rate Models: Theory and Practice, 2nd edition. Heidelberg: Springer Finance.

Brigo, Damiano, and Massimo Morini. 2005. “CDS Market For-mulas and Models.” Unpublished.

Chatterjee, Somnath, and Andreas A. Jobst. 2019. “ Market- implied Systemic Risk and Shadow Capital Adequacy.” Staff Working Paper No. 823, September. Bank of England, London.

Coles, Stuart. 2001. An Introduction to Statistical Modelling of Ex-treme Values. Heidelberg: Springer Verlag.

Crump, Richard K., Stefano Eusepi, and Emanuel Moench. 2018. “The Term Structure of Expectations and Bond Yield.” April. Staff Reports No. 775, Federal Reserve Bank of New York.

De Marco, Filippo, and Marco Macchiavelli. 2016. “The Political Origin of Home Bias: The Case of Europe.” Finance and Eco-nomics Discussion Series 2016-060, Board of Governors of the Federal Reserve System, Washington, DC.

Dell’Ariccia, Giovanni, Caio Ferreira, Nigel Jenkinson, Luc Lae-ven, Alberto Martin, Camelia Minoiu, and Alex Popov. 2018. “Managing the Sovereign- Bank Nexus.” September. IMF De-partmental Paper 18/16, International Monetary Fund, Washington, DC.

Embrechts, Paul, Claudia Klüppelberg, and Thomas Mikosch. 1997. Modeling Extremal Events for Insurance and Finance. Hei-delberg: Springer Verlag.

Enria, Andreas, Adam Farkas, and Lars J. Overby. 2016. “Sover-eign Risk: Black Swans and White Elephants.” European Econ-omy 1: 51–71.

EBA (European Banking Authority). 2010. “Aggregate Outcome of the 2010 EU- wide Stress Test Exercise Coordinated by CEBS in Cooperation with the ECB.” July. European Banking Authority, London.

———. 2011a. “2011 EU- wide Stress Test: Methodological Note.” March. European Banking Authority, London.

———. 2011b. “Capital Buffers for Addressing Market Concerns over Sovereign Exposures– Methodological Note.” October. European Banking Authority, London.

———. 2012. “Final Report on the Implementation of Capital Plans Following the EBA’s 2011 Recommendation on the Cre-ation of Temporary Capital Buffers to Restore Market Confi-dence.” October. European Banking Authority, London.

———. 2014. “2014 EU- wide Stress Test: Aggregate Results.” Oc-tober. European Banking Authority, London.

———. 2016. “2016 EU- wide Stress Test: Results.” July. Euro-pean Banking Authority, London.

———. 2017a. “Annex V. Reporting on Financial Information.” April. European Banking Authority, London.

———. 2017b. “Final Report: Draft Implementing Standards Amending Implementing Regulation (EU) No 680/2014.” April. European Banking Authority, London.

———. 2018a. “2018 EU- wide Stress Test: Methodological Note.” January. European Banking Authority, London.

———. 2018b. “2018 EU- wide Stress Test: Results.” November. European Banking Authority, London.

ECB (European Central Bank). 2011. “Appendix 4—Guidance for Calculations of Losses Due to Application of Market Risk Parameters and Sovereign Haircuts.” March (Frankfurt am Main: European Central Bank).

ESRB (European Systemic Risk Board). 2015. “Report on the Reg-ulatory Treatment of Sovereign Exposures.” March 20. Euro-pean Central Bank, Frankfurt am Main.

FDIC (Federal Deposit Insurance Corporation). 2015. “Regula-tory Capital Rules: Accumulated Other Comprehensive In-come (AOCI) Opt-Out Election.” Financial Institution Letters, March. Federal Deposit Insurance Corporation, Washington, DC.

Fisher, Ronald A., and Leonard H. C. Tippett. 1928. “Limiting Forms of the Frequency Distribution of the Largest or Smallest Member of a Sample.” Proceedings of the Cambridge Philosophi-cal Society 24 (2): 180–90.

Fuster, Andreas, and James Vickery. 2018. “Regulation and Risk Shuffling in Bank Securities Portfolios.” June. Staff Reports No. 851, Federal Reserve Bank of New York.

Gnedenko, Boris V. 1943. “Sur la Distribution Limite du Terme Maximum d’Une Série Aléatoire.” Annals of Mathematics 44 (3): 423–53.



Grandia, Roel, Petra Hänling, Michelina Lo Russo, and Pontus Åberg. 2019. “Availability of High- quality Liquid Assets and Monetary Policy Operations: An Analysis for the Euro Area.” February. Occasional Paper Series No. 218, European Central Bank, Frankfurt am Main.

Gray, Dale F., and Andreas A. Jobst. 2010a. “New Directions in Financial Sector and Sovereign Risk Management.” Journal of Investment Management 8 (1): 23–38.

———. 2010b. Global Financial Stability Report. Chapter 1. Washington, DC, October.

———. 2011a. “Modeling Systemic and Sovereign Risk.” In Les-sons from the Financial Crisis, edited by Arthur Berd. London: RISK Books.

———. 2011b. “Modelling Systemic Financial Sector and Sover-eign Risk.” Sveriges Riksbank Economic Review 2: 68–106.

Hannoun, Hervé. 2011. “Sovereign Risk in Bank Regulation and Supervision: Where Do We Stand?” Paper presented at the High- Level Meeting for the Middle East and North Africa Region, Financial Stability Institute and Arab Monetary Fund, Abu Dhabi, UAE, October 21.

International Monetary Fund (IMF). 2011a. “Germany: Financial Sector Stability Assessment.” IMF Country Report 11/169, June 20, Washington, DC.

———. 2011b. “United Kingdom: Stress Testing the Banking Sector Technical Note.” IMF Country Report 11/222, July 11, Washington, DC.

———. 2011c. “United Kingdom: Financial System Stability As-sessment.” IMF Country Report 11/227, July 1, Washington, DC.

———. 2011d. “Germany: Technical Note on Stress Testing.” IMF Country Report 11/371, December 23, Washington, DC.

———. 2012. “Spain: Financial System Stability Assessment.” IMF Country Report 12/137, December 23, Washington, DC.

———. 2013a. Global Financial Stability Report. Chapter 2. Washington, DC. April.

———. 2013b. “Belgium: Financial System Stability Assessment.” IMF Country Report 13/124, May 17, Washington, DC.

———. 2013c. “Belgium: Technical Note on Stress Testing the Banking and Insurance Sectors.” IMF Country Report 13/137, May 24, Washington, DC.

———. 2013d. “France: Financial Sector Assessment Program-Technical Note on Stress Testing the Banking Sector,” IMF Country Report No. 13/185, June 1, Washington, DC.

———. 2013e. “Italy: Financial Sector Assessment Program-Technical Note on Stress Testing the Banking Sector.” IMF Country Report No. 13/349, December 6, Washington, DC.

———. 2014a. Global Financial Stability Report. Chapter 2. Washington, DC. April.

———. 2014b. “IMF Executive Board Reviews Mandatory Finan-cial Stability Assessments Under the Financial Sector Assess-ment Program.” Press Release 14/08, January 13, International Monetary Fund, Washington, DC.

———. 2014c. “People’s Republic of China–Hong Kong Special Administrative Region: Financial System Stability Assess-ment,” IMF Country Report 14/130, May 22, Washington, DC.

———. 2014d. “People’s Republic of China–Hong Kong Special Administrative Region: Technical Note on Stress Testing the Banking Sector.” IMF Country Report 14/210, July 16, Wash-ington, DC.

———. 2015. “From Banking to Sovereign Stress: Implications for Public Debt.” IMF Policy Paper, Washington, DC.

———. 2017. “Japan: Financial Sector Assessment Program-Tech-nical Note-Systemic Risk Analysis and Stress Testing the Fi-nancial Sector.” IMF Country Report 17/285, September 18, Washington, DC.

Jeanblanc, Monique, and Marek Rutkowski. 2000. “Default Risk and Hazard Process.” In Mathematical Finance—Bachelier Congress 2000, edited by Helyette Geman, Dilip Madan, Stan-ley R. Pliska, and Ton Vorst. Heidelberg: Springer Finance.

Jenkins, Patrick. 2019. “EU’s Wilful Blindness to Sovereign Risk Adds to Eurozone Danger.” Opinion. Financial Times. January. https://www.ft.com/content/c83f64c2-103a-11e9-a3aa-118c761d2745 (subscription required).

Jobst, Andreas A. 2007. “Operational Risk– The Sting Is Still in the Tail but the Poison Depends on the Dose.” Journal of Opera-tional Risk 2 (2): 1–56.

Jobst, Andreas A., Li Lian Ong, and Christian Schmieder. 2013. “An IMF Framework for Macroprudential Bank Solvency Stress Testing: Application to S- 25 and Other G- 20 Country FSAPs.” IMF Working Paper 13/68, International Monetary Fund, Washington, DC.

———. 2017. “Macroprudential Liquidity Stress Testing in FSAPs for Systemically Important Financial Systems.” IMF Working Paper 17/102, International Monetary Fund, Washington, DC.

Jobst, Andreas A., and Hiroko Oura. 2019. “Sovereign Risk in Macroprudential Solvency Stress Testing.” IMF Working Paper 19/266, International Monetary Fund, Washington, DC.

Longstaff, Francis A., Jun Pan, Lasse H. Pedersen, and Ken-neth J. Singleton. 2011. “How Sovereign Is Sovereign Credit Risk?” American Economic Journal: Macroeconomics 3 (2): 75–103.

Merton, Robert C. 1973. “Theory of Rational Option Pricing.” Bell Journal of Economics and Management Science 4 (Spring): 141–83.

———. 1974. “On the Pricing of Corporate Debt: The Risk Struc-ture of Interest Rates.” Journal of Finance 29 (May): 449–70.

Schmitz, W. Stefan, Michael Sigmund, and Laura Valderrama. 2017. “Bank Solvency and Funding Cost: New Data and New Results.” IMF Working Paper 17/116, International Monetary Fund, Washington, DC.

Segoviano, Miguel. 2006. “Portfolio Credit Risk and Macroeco-nomic Shocks: Applications to Stress Testing under Data Re-stricted Environments.” IMF Working Paper 06/283, International Monetary Fund, Washington, DC.

Sondow, Jonathan. 1998. “An Antisymmetric Formula for Euler’s Constant.” Mathematics Magazine 71 (3): 219–220.

Thérond, Pierre, and Pierre Ribereau. 2012. “Théorie des Valeurs Extrêmes.” L’actuariel 5 (June): 52–54.

Vandewalle, Börn, Jan Beirlant, and Mia Hubert. 2004. “A Robust Estimator of the Tail Index Based on an Exponential Regres-sion Model.” In Theory and Applications of Recent Robust Meth-ods (Statistics for Industry and Technology series), edited by Mia Hubert, Greet Pison, Anja Struyf, and Stefan Van Aelst. Basel: Birkhäuser.

Wong, Eric, and Cho- Hoi Hui. 2009. “A Liquidity Risk Stress- Testing Framework with Interaction between Market and Credit Risks.” Working Paper 06/2009, Hong Kong Monetary Authority.

World Bank. 2016. “Doing Business 2017—Equal Opportunity for All.” Washington, DC: World Bank.


https://www.ft.com/content/c83f64c2-103a-11e9-a3aa-118c761d2745

https://www.ft.com/content/c83f64c2-103a-11e9-a3aa-118c761d2745

CHAPTER 10

Revisiting Risk- Weighted Assets: Why Do RWAs Differ across Countries and What Can Be Done about It?

VANESSA LE LESLÉ • ANDREAS A. JOBST

This chapter provides an overview of the concerns about differences in the calculation of risk- weighted assets across banks and jurisdictions and how they might undermine the Basel III capital adequacy framework.1 It discusses potential causes of these differences, drawing upon a sample

of systemically important banks from Europe, North America, and Asia Pacific. The proposed policy recommendations are discussed with a view to remedying actual and perceived problems with risk- weighted assets and improving the use of risk- sensitive capital ratios.

across banks. There has been no convergence in views about the materiality and relative importance of these differences, and, thus, no consensus on policy implications.

From a bank riskmanagement perspective, a clear under-standing of the variations in RWAs is crucial. For example, solvency stress tests ultimately focus on the impact of simulated shocks on capital adequacy ratios. Hence, the im-plications of specific shock scenarios for RWAs— depending on how they are calculated— are key in the interpretation of any stress test results and recommendation of mitigating actions.

In 2012, the Basel Committee on Banking Supervision initiated the Regulatory Consistency Assessment Program (RCAP), which ensures consistent implementation of the Basel framework to help (1) strengthen the resilience of the global banking system, (2) maintain market confidence in regulatory capital ratios, and (3) provide a level playing field for banks operating internationally. The RCAP has resulted

1. INTRODUCTIONStrengthening capital ratios became a priority for the reform of banking regulation in the wake of the global financial cri-sis. However, regulatory reforms have primarily focused on improving the quantity and quality of capital, while changes to the denominator of the capital ratio, that is, risk- weighted assets (RWAs), have been more limited. Confidence in re-ported RWAs was profoundly shaken by the crisis, and supra-national institutions, academics, and market analysts have questioned their reliability and comparability. These concerns risk undermining the relevance of risk- based capital ratios as a cornerstone of international banking regulation (Watt 2011; Fitch 2010, 2012).

Nonetheless, the analysis of RWAs has been surprisingly limited despite the vast academic literature on bank capital. Most research on the characteristics and risk analytics un-derpinning RWAs has been completed by financial analysts, who highlight continued and significant variations in RWAs

This chapter is an updated and substantially revised version of IMF Working Paper 12/90 (Le Leslé and Avramova 2012). The authors are grateful for con-tributions and comments from Jonathan Fiechter, Ceyla Pazarbasioglu, Fabiana Melo, Katharine Seal, Amadou Sy, Anna Ilyina, Ivan Guerra, and John Kiff. The working paper also benefited from discussions with market participants and public sector officials. 1 The working paper on which this chapter is based was originally published when the Basel III framework was still under discussion. While the description

of the rules governing risk-weighted assets (and associated illustrations) have been updated, the empirical analysis reflects historical data that may not fully reflect all aspects of the current framework, which was finalized in December 2017 (BCBS 2017a).


Revisiting Risk-Weighted Assets: Why Do RWAs Differ across Countries and What Can Be Done about It?230

in an analysis of RWAs in the trading book (BCBS 2013a) and two analyses of the RWAs for credit risk in the banking book (BCBS 2013b and 2016c), which focus on the varia-tion in RWAs of banks using internal- ratings- based (IRB) models.2

This chapter explores the scale of variations of RWAs and identifies possible policy actions, consistent with the final-ization of the Basel framework. The revamped framework seeks to restore “credibility in the calibration of RWAs and improve the quality, transparency, and comparability of bank capital ratios” (BCBS 2017a). The chapter (1) discusses the importance of RWAs in the regulatory capital frame-work, (2) highlights the main concerns and the controversy surrounding RWA calculations, (3) identifies key drivers be-hind the differences in RWA calculations across jurisdictions and business models, and (4) concludes with a discussion on the range of options that could be considered to restore con-fidence in banks’ RWA numbers. However, this chapter does not compare the relative merits of leverage and risk- based capital ratios or discuss broader questions, such as (1) better ways to measure risk or predict losses; or (2) the optimal amount of capital banks should hold per unit of risk, which is outside the scope of this study.

2. RISK-WEIGHTED ASSETS, CAPITAL, AND THE REGULATORY FRAMEWORK

RWAs as an Important Component of Capital Ratios

Risk- based versus unweighted capital ratios. Capital ratios are a key indicator of a bank’s solvency and resilience to the re-alization of unexpected losses. Over time, the risk- adjusted

capital framework for banks has changed significantly but remains heavily dependent on RWAs. Until the postcrisis regulatory reforms were introduced, all solvency measures under the Basel framework had long been defined exclu-sively in terms of RWAs.3 The revised Basel framework in-troduced a new solvency measure, the leverage ratio, initially defined as Tier 1 capital over total unweighted on- and off- balance- sheet assets. While Basel III fosters greater conver-gence in the definition and composition of the numerator of capital, the denominator is the product of a mix of Basel ap-proaches, in that it is still subject to the coexistence of vari-ous approaches under the Basel regimes (Figure 10.1).

Why Do We Need to Look at RWAs?

RWAs have at least three important microprudential and macroprudential functions regarding credit, market, and operational risks faced by banks. They (1) provide a credible, comparable, and transparent measure of unexpected losses; (2) ensure the efficient allocation of capital commensurate to associated risks; and (3) serve as indicator of potential asset mispricing (which could signal asset bubbles).

Capital ratios remain the bedrock of the regulatory framework for banks, even after the global financial crisis. Policymakers, banks, and investors all rely heavily on capital ratios to assess the health of banks. Enhancing the mini-mum capital adequacy ratio (together with a more stringent definition of eligible capital) has been the hallmark of the postcrisis reform of Basel II (BCBS 2017a). In 2011, the Eu-ropean Banking Authority (EBA) temporarily raised the risk-based capital requirements in its Capital Exercise of Eu-ropean banks. Similarly, the new capital buffer for global

2 All RCAP reports provide key findings from reviewing banks’ indepen-dent model validation functions, including the methodology and scope of banks’ validation functions and the governance of the validation process.

Sources: Basel Committee for Banking Supervision; and authors.Note: IRB = internal-ratings-based approach; RWAs = risk-weighted assets.

Figure 10.1 The Legacy of Previous Versions of the Basel Framework in Basel III

3 The capital ratios under Basel I/II are based on Tier 1, Tier 2, and total capital, except the total capital ratio. In the wake of the global financial crisis, Basel III replaced these capital ratios with capital measures based on Common Equity Tier 1, and additional Tier 1, and introduced other (temporary) solvency measures, such as Core Tier 1.

Common Equity Tier 1

Additional Tier 1

Total Capital

RWAs

Basel I Basel II

Basel III

Basel III Capital

RWAs

Credit Risk Market Risk Operational Risk

Basel I

Basel II

Basel 2.5

Basel III

Basel IISimplified Standardized Foundation

IRBAdvanced

IRBBasel III


Vanessa Le Leslé and Andreas A. Jobst 231

RWAs, which, as an extreme example, has resulted in a com-plete removal of an internal modeling option, such as in the case of operational risk (BCBS 2016b).

This chapter aims to contribute to the debate on RWAs. Starting with the premise that retaining risk- based capital ratios is the preferred outcome, the chapter maps out con-cerns and differences and suggests possible policy options to strengthen the current RWA framework.

3. WHAT ARE THE KEY CONCERNS ABOUT RWAS?The global financial crisis raised doubts about the adequacy, consistency, and comparability of regulatory capital ratios (summarized in Table 10.1), with concerns focusing primar-ily on the reliability of reported RWAs. Given the fundamen-tal role of RWAs in risk- adjusted capital frameworks, this could cause market participants to (1) adjust capital ratios (likely downward); (2) replace capital ratios with leverage ratios in their capital assessment; (3) require higher capital ratios to compensate for the low perceived reliability of RWAs; and/or (4) restrict lending to banks, for which they have doubts about reported capital adequacy.

Comparing Capital Ratios

Capital ratios are fundamental to a comparative assessment of bank solvency. However, the findings in this chapter sug-gest that, in the transition toward the finalization of the Basel framework until the end of 2017, (1) the capital levels vary greatly depending on the definition of the capital ade-quacy measure used (that is, either risk- based or unweighted); (2) headline capital ratios may mask very different risk levels, or at least different measurement approaches; and (3) banks converge toward the regulatory capital ratio that is the most favorable to them. In this chapter, the discussion is based on a sample of 50 systemically important banks (SIB)4 based in three regions: Asia Pacific (“Asia” color- coded in yellow), Europe (“EU” color- coded in blue), and North America (“NA” color-coded in red). The sample is also broken down into three main (simplified) business models: retail (com-mercial) banks, investment banks, and universal banks.

Banks’ capital adequacy varies significantly, depending on whether capital ratios are risk weighted or not. Figure 10.2 compares the dispersion in capitalization levels by regions (Eu-rope, North America, and Asia Pacific), based on three indica-tors: (1) Core Tier 1 capital5 over RWAs, (2) tangible common equity over tangible total assets, and (3) tangible common eq-

systemically important banks is based on a percentage of RWAs. However, increased reliance on capital ratios comes at a time when their robustness is being questioned, as vari-ous capital measures provide different and often conflicting messages about banks’ solvency.

Capital ratios have been scrutinized for as long as they have been in existence, but the focus has only gradually shifted from the numerator to the denominator. The per-ceived differences in the detailed application of the existing Basel standards for the determination of RWAs have raised concerns about the credibility and effectiveness of the capital framework. As early as 1999, the Basel Committee noted that “with increasing sophistication of the banks and the devel-opment of new innovative techniques in the market, the largest banks have started to find ways of avoiding the limitation which fixed capital requirements place on their risk- taking relative to their capital. For certain banks, this is undoubtedly starting to undermine the comparability and even the meaningful-ness of the capital ratios maintained” (BCBS 1999). This con-cern has mostly manifested in efforts aimed at introducing greater granularity and sophistication in the definition of capital adequacy prior to the financial crisis; however, exces-sive risk taking and skewed incentives due to the low capital intensity of banks’ trading book and off- balance- sheet expo-sures were identified as two of the accelerants of the financial crisis, and, thus, received much attention in the postcrisis regulatory reforms.

Regulatory changes have responded to the evolving mar-ket perception of the reliability of the capital ratio to ensure that capital adequacy around the world is implemented us-ing consistent metrics of banks’ capital and risks. While in-vestors are now mostly concerned about the denominator of capital ratios, this was not true in the run- up to the global financial crisis when there was a general preference for alter-native capital measures that would better reflect the loss- absorbing capacity of capital (for example, Core Tier 1 in Europe and Tangible Common Equity or Tier 1 Common in the United States). Basel III then adopted a much stricter definition of capital to correct the main deficiencies of the numerator. In the wake of the financial crisis, rising mistrust in the way certain banks calculate their RWAs (particularly the ones using the Basel II advanced IRB approach) has re-sulted in calls for a higher minimum capital ratio to com-pensate for the possible understatement of RWAs and a rising preference for unweighted capital measures (leverage ratio) as an alternative (or complementary) way of assessing capital adequacy. Either way, this underlines the urgency to revisit RWAs.

There are still practical limitations for global comparisons of regulatory capital ratios owing to national heterogene-ities and varying acceptance of internal models. Also lengthy transitional arrangements, including the grandfathering by regulators of certain hybrid debts until 2023, may continue to blur comparability and consistency over time of banks’ Tier 1 ratios. There has also been a noticeable shift in the regulatory stance toward a more critical review of internal models for

4 A complete presentation of the sample and methodology is available in Appendix 10.2. Analysis solely relies on publicly available data, which frequently lack full disclosure and consistency.

5 The Tier 1 capital ratio is a more widely reported indicator, but the numerator varies significantly across jurisdictions and is not the stron-gest measure to assess capital adequacy.



TABLE 10.1

Overview of Key Concerns about RWA Calculation PracticesKey Concerns Possible Impact

Regulator Perspective

Reliability and accuracy of capital ratios• Inaccurate risk measurement (both on- and off-

balance sheet).• Understatement of risk.• Mispricing of low- probability, high- impact events

(“tail risk”).

• Overstatement of capital adequacy, which conveys: Inaccurate assessment of solvency and resilience. Unreliable indicator of stress, possibly delaying

timely resolution decision.

Adequacy of capital• RWAs have declined during and after the global

financial crisis (despite a heightened risk environment).

• The optimization of risk models, data cleaning, and parameter updates seems to decrease RWAs.

• Banks with similar business models may have very different RWAs for similar exposures.

• Inconsistencies between risk exposure and capital require-ments tend to inflate capital ratios.

• Some banks might underestimate unexpected losses.• Lower risk weighting suggests greater scope for error in the

calculation of capital requirements.

Procyclicality• RWAs based on ratings (and/or using historical

parameters) may be too low in good times and rise too late during times of stress.

• The way the default probability (“point in time” vs. “through the cycle”) is estimated affects the cyclicality of capital.

• Calculation of RWA may amplify the procyclicality of capital requirements, as banks would try to reduce their RWAs during the down cycle, which might encourage greater risk taking during recovery.

Risk- taking incentives and risk management• Banks may deliberately underestimate risks to reduce

their capital intensity.• RoE- type profitability targets may incentivize banks

to reduce RWAs as much as possible.

• Lack of sufficient oversight and excessive management discretion in reducing capital intensity may result in excessive risk taking and could potentially lead to bank failure, with significant related social and economic costs.

Bank Perspective

Competitive advantage• Banks with lower RWAs (despite comparable risk)

could benefit from an undue competitive advantage (due to lower capital requirements).

• The capital add- on for GSIBs is calculated as a percentage of RWAs (not of total assets), which favors banks with structurally lower risk weightings.

• Least conservative banks could capture higher market share, which could threaten global financial stability.

• The GSIB capital add- on does not necessarily penalize the largest banks (in terms of total assets).

• Cross- country variation of: Regulations and supervision of banks’ RWA

practices, and Review and robustness of validation and approval

of internal models for RWAs.

• Inconsistent treatment makes the application of Pillar II and capital buffers more variable and, thus, less comparable.

• Some banks might be subject to less scrutiny due to varying supervisory practices across countries.

Market Perspective

Comparability of capital ratios• RWAs are subjective and vary across banks, which

complicates comparability within countries and across national boundaries.

• Markets may prefer a simpler, more objective, and easier- to- compare measure such as the leverage ratio.

Credibility of capital ratios• Different methodologies may cause banks, regula-

tors, and markets to distrust each other on reported RWAs.

• Uncertain RWAs could lead to loss of confidence in regulatory actual bank solvency, where markets become reluctant to lend to banks, ultimately resulting in a liquidity crisis.

Opacity and complexity of internal models• Internal modeling approaches for RWAs can be very

complex, leaving considerable room for interpretation.

• The quality of internal models (and the robustness of associated methodologies) might be difficult to assess.

• Markets may doubt the quality of the capital ratio and switch to the leverage ratio as a basis for assessing capital adequacy.

• Regulators may be tempted to override internal models and impose minimum floors for risk weights.

• Erosion of trust in a risk- based framework (Basel III) in favor of a less risk- sensitive solvency regime (Basel I).

Source: Authors.Note: GSIB = global systemically important bank; IRB = internal-ratings- based; RoE = return on equity; RWA = risk- weighted asset.



77 percent (North America). Both panels show large gaps between the lowest and highest leverage ratios and RWA densities reported within each region, suggesting that the same reported Core Tier 1 ratio may mask differences in the level of risks it supports across banks. In fact, the European case highlights that banks prefer a risk- based capital ratio if the actual RWA density remains below the RWA density im-plied by the leverage ratio (that is, about 35 percent for a le-verage ratio of 3.0 percent if the minimum capital requirement (in the form of Common Equity Tier 1) is 8.5 percent). However, it may also be driven by a combina-tion of different types of assets and associated risk weights across sample banks.

Linking ratings and capital measures also highlights the difficulty in relying on any single measure of capitalization to assess a bank’s solvency. For the broader sample of 50 banks, Figure 10.4 shows a very wide dispersion of re-ported Core Tier 1 (or equivalent) for banks rated in the same category. For instance, “AA”-rated banks (based on an average of Standard & Poor’s and Moody’s credit ratings) report a Core Tier 1 ratio of between 9 percent and 21 per-cent. Similarly, the RWA density varies considerably within each rating bucket, suggesting that ratings are not a strong indicator per se of bank solvency, as they incorporate other quantitative and qualitative parameters. For the narrower sample of 14 banks, ratings also point to a variation in the overall quality assessment of banks.

uity over RWAs. All three indicators point to higher capital levels in Asia Pacific and North America, with wider dispersion among European banks. The latter are scattered when ranked in terms of risk- based capital ratios (that is, Core Tier 1 or tangible common equity/RWAs), but drop to the bottom of the sample list when assessed on an unweighted basis (leverage ratio), while US and Asia Pacific banks rise to the top, indicat-ing that the average RWA density (that is, RWAs in percent of total assets) of the former is significantly lower.

The better performance of banks in certain geographical lo-cations under certain capital ratios is driven by a combination of factors, including (1) the regulatory environment, (2) the ac-counting framework, (3) the economic cycle and its impact on the estimation of default probabilities; and (4) firm-specific dif-ferences in business models, the composition of RWAs (credit/market/operational risk), and RWA methodologies.

Conversely, similar capital ratios may reflect very differ-ent levels of risk due to these factors (see Section 4: Key Drivers of Differences in RWA Calculations). For instance, for a subsample of 14 banks from Europe, North America, and Asia Pacific, which reported the same Core Tier 1 ratio (or equivalent) of 9 percent, the corresponding leverage ratio and RWA density varied significantly, both across and within regions (see Figure 10.3). The leverage ratio con-verged at around 3 percent in Europe and Asia Pacific, less than half the value for banks in North America. Similarly, for RWA density, the level goes from 23 percent (Europe) to

Asia Pacific Europe North America

Global Average11.0%

EU 10.6%NA 10.5%AP 11.9%

3. Core Tier 1 Equivalent (TCE/RWA)

- 5 10 15 20 - 3 5 8 10 - 3 5 8 10 13 15 18 20 23

Global Average4.9%

EU 3.5%NA 5.9%AP 5.8%

Global Average11.0%

1. Core Tier 1 to Risk-Weighted Assets

2. Tangible Common Equity to Tangible Total Assets

Sources: Bloomberg LP; and authors’ calculations.Note: AP = Asia Pacific; EU = Europe; NA = North America; RWAs = risk-weighted assets; TCE = tangible common equity.

Figure 10.2 Core Tier 1 Capital Ratio, Leverage Ratio, and Core Tier 1 Equivalent, as of June 2011 (In percent)



many banks grew their balance sheets. However, limited infor-mation makes it difficult to explain whether total assets and RWAs are moving in step or diverging. The evidence suggests that RWAs have been declining proportionally more than have total assets for some banks and jurisdictions. Several factors may explain this decrease in RWA density. As intended by de-sign, the gradual shift from Basel I to Basel II (particularly IRB approaches) has enabled banks to benefit to some extent from lower RWAs, as they move their portfolios to the advanced IRB approach, which reduces the amount of unexpected losses reflected in risk weights.

The decline of average RWAs could also have been caused by changes in the business mix and portfolio rebalancing, with banks shifting to those assets that carry lower risk weights while reducing exposures attracting higher capital requirements. In addition, it is also important to consider the point in the cycle (growth or downturn period) at which the probability of default (PD) and risk weights are calculated.

Variability of RWAs and Total Asset Variations

Total assets, RWAs and RWA density vary across jurisdic-tions and over time. The RWA density seems to be a good indicator of a bank’s riskiness in theory. However, there are significant differences between average RWA densities across regions (Figure 10.5), prompting questions about consistency of different RWA methodologies used by banks, and, thus, casting doubts on the reliability of banks’ capital ratios. In Europe, the RWA density dropped dramatically in 2001 and has been trending downward ever since (especially after banks had transitioned to Basel II by 2007) as European SIBs (which mostly follow a “universal banking model”) diversi-fied their business lines and geographies; in contrast, the av-erage RWA density of US banks (where banks reported under Basel I for much longer than in Europe) was about 50 percent higher than that of European banks.

Generally, RWA densities had started decreasing (or stopped increasing) several years prior to the global financial crisis as

0

8

1

5

6

7

3

4

2

1. Dispersion of Leverage Ratio with Core Tier 1 Ratio at 9 Percent

0

9080

10

506070

3040

20

2. Density of RWA by Region with Core Tier 1 Ratio at 9 Percent

Europe(seven banks)

Asia Pacific(two banks)

North America(five banks)

Europe(seven banks)

Asia Pacific(two banks)

North America(five banks)

4

6

7

3 34

46

57

77

38

23

40

Sources: Bloomberg, LP; SNL; and authors’ estimates.1Sample of 14 global banks with Core Tier 1 ratios of 9 percent.

Figure 10.3 Leverage Ratio and RWA Density for a Global Sample of Banks with the Same Core Tier 1 Ratio, as of 20111 (In percent)

Sources: Bloomberg, LP; SNL; Standard & Poor’s; Moody’s; and authors’ calculations.Note: CT1 = Core Tier 1; RWA = risk-weighted assets.1Only 14 sample banks with the same Core Tier 1 capital ratio of 9 percent.

Figure 10.4 CT1 Ratios and RWA Density by Ratings for Sample Banks, as of 20111

1214

20

468

10

1618

2224

1. Core Tier 1 Ratio or Equivalent Range per Credit Rating (Percent)

AA AA– A+ A A–

BBB+

BBB–

21

14 1416

18

9

14

9

79

810

7

4050

80

0102030

6070

90

2. Density of RWA over Total Assets per Credit Rating (Percent)

AA AA– A+ A A–

BBB+

BBB–

4

0

1

2

3

5

3. Number of Banks by Credit Rating at a 9 Percent Core Tier 1 Ratio (OriginalCredit Rating)

AA AA– A+ A

BBB+

47

79

57 57

8276

80

33

1726

17

34

62

74



ing some additions and amendments to Basel II), which differ significantly (see Appendix 10.1). Even within Basel II, there are three possible approaches to choose from: the standardized approach, the foundation internal- ratings- based (FIRB) ap-proach, and the advanced internal- ratings- based (AIRB) approach. Basel II imposes capital charges for operational risk, whereas Basel I does not.

Banks in most systemically important jurisdictions had reported under Basel II6 (Figure 10.6), with the AIRB ap-proach being the most commonly used (14 countries). In Europe, all banks were required by European legislation— Capital Requirements Regulation and Capital Requirements Directive (CRD) (European Union 2013a, 2013b)—to im-plement Basel II in 2007. In Asia, countries with large finan-cial sectors followed one of the Basel II approaches as well. In contrast, the implementation of Basel II was protracted in the United States, with new risk- sensitive and granular capi-tal rules coming into effect only in 2012 (OCC/Federal Reserve/FDIC 2012). The analysis and data in this chapter show US banks reporting under Basel I (until the end of 2011), with the largest internationally active US banks (with total assets of over $250 billion) in a parallel- run phase, with a view to migrating to Basel II (and III).

Banks have the incentive to optimize capital requirements across assets with different risk weights if the regulatory framework becomes more risk sensitive. Banks would select assets that look attractive under their regulatory regime. For instance, European banks reported— on average— a persis-tently lower average RWA density than banks in other regions, suggesting a general preference for assets that carry a low risk weight (assuming that banks tended to have similar risk expo-sures); this allowed them to report strong capital ratios under

Deteriorating macroeconomic conditions should generally increase PDs (and, thus, raise risk weights); however, greater use of collateral (which decreases the loss given default) may have contributed to a decline in the RWA density on average. Some banks (particularly advanced IRB banks) may have amplified this decline by changing their RWA methodology further to obtain lower RWAs.

4. KEY DRIVERS OF DIFFERENCES IN RWA CALCULATIONSA host of factors influence banks’ risk- modeling choices. Some reflect true risk taking and are bank- specific parame-ters (such as the business model or asset quality), while oth-ers are mostly external factors, which are unrelated to the risks that banks take on (such as institutional, accounting, and regulatory parameters). These differences are plausible and expected, as operating environments and business mod-els vary widely among banks. This section aims to identify the causes for these variations and quantify their impact while differentiating between bank- specific and external pa-rameters across regions and types of banks.

Overview of Factors Influencing RWAs

Table 10.2 provides a simplified overview of the main factors behind RWA differences, which are detailed in the following paragraphs.

External Parameters

Regulatory frameworks are a key factor influencing the calculation of RWAs

RWAs are primarily driven by local regulatory requirements. Banks follow either the Basel I or Basel II framework (and/or had already anticipated the finalization of Basel III by adopt-

Asia Pacific Europe North America

0

100

10

50

60

80

90

70

30

40

20

Dec-

98

Jun-

99

Dec-

99

Jun-

00

Dec-

00

Jun-

01

Dec-

01

Jun-

02

Dec-

02

Jun-

03

Dec-

03

Jun-

04

Dec-

04

Jun-

05

Dec-

05

Jun-

06

Dec-

06

Jun-

07

Dec-

07

Jun-

08

Dec-

08

Jun-

09

Dec-

09

Jun-

10

Dec-

10

Jun-

11

57%

35%

51%

Sources: Bloomberg, LP; SNL; and individual bank reports.1The sample is not constant over the period (due to consolidation and changes to the perimeter of banks under review), comprising eight banks at its minimum (1998) and 50 from 2008 onward. In addition, the sample masks considerable in-sample variation, with some banks experiencing a secular increase in risk-weighted assets while others see their risk-weighted assets decline.

Figure 10.5 Evolution of Risk-Weighted Asset Density (1998–2011)1 (In percent)

6 This classification is simplified, as some banks may have opted for a dif-ferent approach from the one that is the most common in their country of incorporation, and many banks use a combination of approaches.



TABLE 10.2

Main Factors of Differences in RWA Densities across Jurisdictions and BanksExternal Parameters Banks’ Internal Parameters

Regulatory Framework• Different ways in determining risk weights for

different exposures: standardized approach (including the simplified version) vs. foundation and advanced internal- ratings- based (IRB) approaches.

• Cross- country variation in regulatory emphasis on either risk- weighted capital ratios or ( unweighted) leverage measures.

• Banks’ choice in determining risk weights for different exposures, including the combination of approaches for different portfolios and/or geographies.

Supervisory Framework• Initial model validation and ongoing supervision

(including RWA classification methodology) RWA classification methodology.

• Imposition (or not) of minimum prudential floors for PD/LGD.

• Intensity of use of Pillar 2.• Understanding of broader risk management.

• Modeling risk.• Risk management and strategy.• Risk appetite.

Accounting Framework• IFRS versus US GAAP or other local accounting

standards.• Some banks report under both IFRS and US GAAP.• Full or partial implementation of IFRS standards.

Legal Framework• Solvency regime and contract enforcement (“rule of

law”).• Recovery process and collateral access.

• Internal risk management and recovery procedures.

• Collateral use (including encumbrance).

Economic Cycle• Economic growth vs. downturn (mild/severe)

scenarios.• Different default and recovery rates by country asset

class.• Asset classes.

• Asset mix (by geography and business lines).• Lending and recovery practices.• Internal assumptions about PD/LGD and expected

recovery.• PD determined based on either “point in time” or

“through the cycle”; LGD calibrated to “down-cycle” experience.

Business Model• Constraints on bank structure (legal separation of

activities, ring- fencing).• Business model choice: universal bank/retail bank/

investment bank (or combination).• Asset composition and business mix.

Lending, Valuation, and Provisioning Practices• Directed lending (from the government or related

public bodies) or free market.• Mortgages: originate and distribute on- or off-

balance sheet (for example, passed over to GSEs).• Structure of the economy (for example, SMEs versus

large corporates; level of indebtedness of corporate and households).

• Lending practices (loans versus bonds; maturity of assets; quality of borrowers).

• Geographic footprint may impose local practices in addition to those applicable on a group-wide basis.

• Valuation and collateralization of assets.• Classification of assets into performing/nonper-

forming loans.• Provisioning practices.

Source: Authors.Note: GAAP = Generally Accepted Accounting Principles; GSEs = government- sponsored entities; IFRS = International Finan-cial Reporting Standards; IRB = internal- ratings- based; LGD = loss given default; PD = probability of default; RWAs = risk- weighted assets; SMEs = small and medium- sized enterprises.

Banks operating under Basel I report much higher RWA densities than peers that determine their risk weights under the IRB approaches of Basel II (and III). In fact, the average RWA density under the AIRB approach is almost one third lower than the average RWA density under Basel I. This is consistent with the Basel II objective of encouraging banks to develop more advanced riskmanagement techniques (and be “rewarded” with lower RWAs and capital requirements). As banks migrate to more sophisticated approaches, their

the Basel II risk- weighted framework owing to higher leverage (at the expense of lower returns). Conversely, in the United States, where the regulatory emphasis had long been on the leverage ratio, banks traditionally focused on higher- yielding (and riskier) assets that carried attractive returns to compen-sate for a more binding leverage constraint. The empirical evi-dence shows that greater risk sensitivity to the regulatory framework generally resulted in lower RWA density, especially as banks transitioned from Basel I to Basel II (Figure 10.7).



government- sponsored entities, which is not possible in Europe, to reduce their total assets and RWAs. Neither regime is better or worse as such, but they are a result of distinct social and eco-nomic choices. Therefore, comparisons tend to be more mean-ingful when done on a regional basis among banks with similar regulatory setups as well as relatively similar business models.

Prudential floors have helped to mitigate model risk and measurement error stemming from internally modeled ap-proaches. In the transition to full implementation, the Basel Committee introduced a capital floor to ensure banks do not

RWA density is likely to decrease. The use of the FIRB ap-proach is mostly a stepping stone toward the AIRB approach, and, thus, is less pervasive. Most European banks adopted the AIRB approach, but to varying degrees (some almost ex-clusively; others still use a combination of approaches on dif-ferent segments of their portfolios).

Also, institutional differences influence the way banks mea-sure risks and capital. For instance, US and European banks operated under a different institutional setup. In the United States, banks can offload most of their mortgage loans to

Latin AmericaNorth AmericaEuropeAsia Pacific

Basel II vs. Basel I: Number of countriesreporting by region

Basel II (AIRB)14 countriesBasel II (FIRB)

6 countries

Basel I2 countries

Basel II (SA)2 countries

0

16

2

10

12

14

6

8

4

Basel II (AIRB) Basel II (FIRB) Basel II (SA) Basel I

AustraliaHK

JapanSingapore

ChinaS. Korea India US

AustriaDenmarkFrance

ItalyNetherlands

NorwaySpain

SwitzerlandUK Belgium

GermanyIrelandSweden

Canada

MexicoBrazil

Sources: Banks’ Pillar III Reports; and annual reports.Note: AIRB = advanced internal- ratings- based; FIRB = foundation internal- ratings- based; HK = Hong Kong SAR; IRB = internal- ratings- based; SA = standardized approach.

Figure 10.6 Regulatory Frameworks in 24 Jurisdictions with Systemically Important Financial Systems, as of 2011

SA Average 62.9%

Basel I Average62.7%

AIRB Average 38.8%

FIRB Average 44.2%

Basel I

SA

FIRB

AIRB

0 20 40 60 80 100

Base

l II

Source: Authors.Note: AIRB = advanced internal-ratings-based; FIRB = foundation internal-ratings-based; SA = standardized approach.

Figure 10.7 RWA Density by Regulatory Standards, as of June 2011



pean Union 2013a). In the United States, the implementation measures for the Dodd- Frank Wall Street Reform and Con-sumer Protection Act of 2010 (or “ Dodd- Frank Act”) created a permanent Basel I floor on minimum risk- based capital require-ments, which limited incentives for large international US banks to optimize RWA densities as they migrated to Basel II.

The accounting framework matters

Also, the accounting regime influences RWA density. Al-most all jurisdictions with systemically important financial sectors (except China and the United States) permit or re-quire that domestic companies use accounting rules accord-ing to International Financial Reporting Standards (IFRS) (Figure 10.8, panel 1). Globally, 144 jurisdictions require IFRS for all or most companies (IFRS 2018). The European Union has adopted virtually all IFRS standards (with some carve- outs for International Accounting Standard 39 [IAS 39] and its replacement, IFRS 9), and most countries in Asia and the Americas have IFRS or equivalent standards in place. In the United States, the US Generally Accepted Ac-counting Principles (GAAP) remain the prevalent account-ing framework, but work is under way to allow greater convergence toward IFRS. Among the key differences be-tween accounting standards is the netting of derivative

hold insufficient regulatory capital. The capital floor is based on the application of Basel I. It is derived by applying an ad-justment factor to the following amount: (1) 8 percent of RWAs, (2) plus Tier 1 and Tier 2 deductions, and (3) less the amount of general provisions that may be recognized in Tier 2. If the floor amount is larger, banks are required to add 12.5 times the difference to RWAs (assuming a 100 percent risk weight for the gap at a minimum capital requirement of 8 percent). The floor gradually decreased to 80 percent in 2009, when it was supposed to be removed (BCBS 2006). However, the continuing effects of the global financial crisis led the Ba-sel Committee to extend the Basel I floor until further notice (BCBS 2014), which limits the extent to which divergent RWA densities can be explained by differences in regulatory regimes. A floor was also implemented as part of the Euro-pean Banking Authority’s 2011 recapitalization exercise to avert “excessive model optimization” of European banks us-ing IRB approaches for their capital assessment.

However, floors cannot prevent some divergences in imple-mentation across jurisdictions. In Europe, the floor required by the original CRD also expired at the end of 2009, but CRD III reinstated it until the end of 2011. During the transition to CRD IV, which came into force in July 2013 (for transposition into national law until January 2014), national authorities were able to waive the requirement under strict conditions (Euro-

Latin AmericaNorth AmericaEuropeAsia Pacific

0

20

2

10

12

16

18

14

6

8

4

IFRS RequiredAll

Some IFRSPermitted

No IFRS

All IFRSRequired: 17

countries

No IFRS: 4countries

Some IFRSPermitted:3 countries

AustraliaHK

S KoreaJapanChina

IndiaSingapore

AustriaBelgiumDenmarkFrance

GermanyItaly

IrelandNetherlands

NorwaySpain

SwedenUK

Switzerland

Canada

US

Brazil

Mexico

RWA Density by Accounting Standard

0 50 100

IFRS

IFRS*

No IFRS

IFRS* (Permitted)34.6%

No IFRS64.3.0%

IFRS 39.6%

Sources: Banks’ Pillar III reports and annual statements, 2010; and Deloitte & Touche, 2011.Note: HK = Hong Kong SAR; IFRS = International Financial Reporting Standards.

Figure 10.8 Accounting Standards in Sample Countries with Systemically Important Financial Sectors, as of 2011



Grunspan 2011), the estimated default rate for US mortgage loans increased from 20 basis points before 2008 to 200 basis points in the wake of the global financial crisis.

• Default history: Historical rates of default in Europe from 1996 to 2010 were consistently below those in Asia and the United States (Figure 10.9). Defaults in the United States were more elevated than those in Europe for both bonds and loans. So, the differ-ences in default history may have justified the use of lower risk weights by European banks in the past.

• Default risk model: Credit rating agencies differ on how they estimate default rates and the way they in-form their credit assessment. Moody’s and S&P, for instance, report different default rates for the same regions (Figure 10.9).

Bank- Related Parameters

The business model and geographic footprint of banks influ-ence the calculation of RWAs. Banks can be broadly categorized into three groups: (1) retail/commercial banks, (2) universal banks, and (3) investment banks. Investment banks are mostly exposed to market risk, and retail banks are mostly exposed to credit risk. Universal banks cover all activities and are exposed to both credit and market risks.7

The prevalence of certain business models explains differ-ences across and within regions. RWA densities of European banks tend to be lower than those of Asian and North American peers (Figure 10.10); however, general regional differences betray some notable (but simplified) cross- country differences. In Europe, some banks in Spain, Italy, and the United Kingdom, which are more geared toward retail activities, have a higher RWA density than those based in France, Germany, and Switzerland, where universal or investment banking tends to dominate. In North America, the RWA density of US regional banks is higher than that of

positions (authorized under US GAAP, but disallowed under IFRS). Hence, the off- balance- sheet positions would appear more “inflated” on an IFRS basis.

The biggest impact of accounting is likely to be on assets, the denominator of RWA density, rather than on risk weights, the numerator. Since average risk weights for sample banks that do not report under IFRS are, on average, higher than those reported by IFRS banks, the accounting frame-work seems to matter when comparing average RWA densi-ties (Figure 10.8, panels 2 and 3).

Economic cycle and PD assumptions are important in determining RWAs

The PD is a key input for the calculation of risk weights un-der the IRB approaches (BCBS 2017a). Risk weights increase as the PD rises, but the relation is not proportional— a high increase in the PD will typically translate into a more moder-ate increase in the risk weight based on the asymptotic single- risk- factor model underpinning the specification of IRB approaches (Jobst and Weber 2016). For a given loss given default (LGD), risk weights can be derived for different levels of PDs. For example, assuming an LGD of 50 percent, a PD of six basis points generates a risk weighting of 44 percent; however, tripling the PD to 18 basis points increases the risk weight by “only” slightly more than two thirds (75 percent) (Samuels and Harrison 2011a and 2011b).

Several factors can explain the variation of PDs for simi-lar exposures:

• Timing: PDs can be derived as “ point- in- time” esti-mates (that is, using recession periods with elevated default rates) or as “ through- the- cycle” default rates. That, in part, explains why some banks experienced significant RWA inflation as the recession deepened while others did not.

• Reference period: Differences in PD estimates can also arise because of different time periods used in the estimation. For instance, according to the credit rating agency Standard & Poor’s (de Longevialle and

Source: Standard & Poor’s (S&P); and Moody’s.

Figure 10.9 Default Rates by Region and Rating Agency (1996–2010) (In percent)

EU (Moody’s)North America (Moody’s)Europe (S&P Ratings)North America (S&P Ratings)

Global EuropeNorth AmericaAsia Pacific

0

16

2

10

12

14

6

8

4

1. By Region

1996 97 98 99

2000 01 02 03 04 05 06 07 08 09 10

0

8

1

5

6

7

3

4

2

2. By Rating Agency

1996 97 98 99

2000 01 02 03 04 05 06 07 08 09 10

7 Note that operational risk is only accounted for under Basel II.



collateral and has a greater impact on risk weights than PDs (due to the linear relation between risk weights and LGD in the IRB formula). Thus, banks have a regulatory preference for secured lending with a high degree of collateralization. All else equal, this may bias their lending choice toward mort-gages and commercial real estate, which implies higher recov-ery rates, than unsecured retail or corporate loans. Banks might also want to minimize the maturity of exposures. Longer- dated assets attract higher risk weights than assets with a short maturity to account for the greater uncertainty over longer time horizons. Banks whose portfolios have a lon-ger average maturity (for example, large mortgage books) tend to have higher risk weights on average.

Credit risk represents by far the largest component of RWAs and accounts for 86 percent on average for the study sample (Figure 10.11). In aggregate, market and operational risks are broadly equal at 6.5 percent and 7.5 percent, respec-tively, but there are significant regional differences. US banks did not report under Basel II during the study period, and, thus, did not disclose any operational risk. Market risk is concentrated in large global investment banks (and a few universal banks), mostly US and European, whose average is 17 percent. The variation of credit and market RWAs is simi-lar at 38.4 and 36.7 percent (between minima and maxima) across all business models and regions.

The risk weights for credit risk under IRB approaches vary considerably across banks and portfolios. The complex for-mula for calculating RWAs under Basel II allows for different

international and regional banks in Europe and Asia Pacific, mainly because of their mortgage and retail focus. However, US global money- center banks have RWA densities that are below their regional average. Canadian banks exhibit low RWA densities, primarily because parts of the mortgage book are also government guaranteed. In Asia Pacific, some Australian banks, whose business profiles are closer to those of European universal banks than to emerging market Asian banks, rank generally below the regional average.

Retail banks tend to have higher RWA densities than universal banks and investment banks. The relatively lower RWA densities of investment banks are to be expected, given that these banks have large trading books, which attract lower risk weights than banking- book assets. With the revi-sion of the trading book, risk weights for market risk have increased (for example, for securitization, proprietary trad-ing, and derivatives through the credit valuation adjustment) in what is commonly referred to the implementation of “Ba-sel 2.5” (an interim phase prior to the finalization of postcri-sis regulatory reforms in Basel III in 2017).8

In addition, the characteristics of banks’ portfolios heavily influence the calculation of RWAs. The recovery assumption (expressed in LGD) depends on the quantity and quality of

0 50 100 0 50 100 0 40 6020 80

NorthAmerica

EU

AsiaPacific

IB

RB

UB

Average49.6%

1. Density Ratio by Region 2. Density Ratio by Business Model

3. Density Ratio under Basel II AIRB Approach

Average34.4%

Average44.3%

Average57.1%

Average 29.6%

UB

IB

RB

Average40.1%

Average 18.3%Average57.0%

Average45.0%

Source: Bloomberg, LP; individual bank reports, and authors’ estimates.Note: AIRB = advanced internal- ratings- based; IB = investment bank; RB = retail bank; UB = universal bank.

Figure 10.10 RWA Densities for All Banks in the Sample Grouped by Region and by Business Model, as of June 2011

8 Quantitative impact studies by the Basel Committee indicate that the market risk weights are bound to triple on average, with some banks expected to see the RWA density increase over 10 times with the imple-mentation of Basel 2.5.



most credit risk categories, except for interbank lend-ing; this is mostly attributable to US banks, which at the time had not transitioned from the very broad treat-ment of credit risk under Basel I to a more granular and risk- sensitive Basel II framework.10 This is particularly true for unsecured loans, which attract the highest risk weights in North America (up to 100 percent).

• Risk weights of banks in Asia Pacific and Europe are similar for all creditrisk categories. For instance, the risk weights of mortgages (15 percent and 14 percent average, respectively) are much lower than in North America (where banks report a risk weight of 40 per-cent on average). In North America, the risk weights in mortgage portfolios range from 6 percent to 50 percent owing to significant differences between Canada and the United States. While most US banks report risk weights close to 50 percent, government guarantees

estimations of default risk and recovery rates, and, thus, lim-its consistency and comparability.9 The assessment of credit risk remains essentially unchanged under Basel III. For in-stance, research by investment bank analysts on large Euro-pean banks, which rely mostly on IRB approaches, suggests a significant variation of risk weights within each creditrisk category (residential/commercial mortgages, corporates, in-stitutions, and other unsecured retail) (Table 10.3). This might be explained by divergences in the implementation of the Basel II framework by domestic supervisors, such as risk- parameter floors, treatment of nonperforming loans, param-eters for cycle adjustment, and migration matrices.

In general, credit risk weights also vary significantly across credit risk categories in other regions (Figure 10.12):

• North American banks show, on average, higher risk weights (with the highest dispersion across banks) for

Sources: Individual bank reports; SNL; and authors’ estimates.Note: The breakdown of RWAs was obtained from individual annual reports, which might cause inconsistencies in the definition and delimitation of risks. RWA = risk-weighted assets.

Figure 10.11 Breakdown of RWAs by Credit, Market, and Operational Risks, as of June 2011 (In percent)

102030405060708090

Operational Market Credit

0

100

N. A

mer

ica

N. A

mer

ica

N. A

mer

ica

N. A

mer

ica

N. A

mer

ica

Asia

Asia

Asia

Asia

Asia

Asia

Asia

Asia

Asia

Asia

Asia

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

9 Andy Haldane (2011) observed that the “number of risk buckets has increased from around seven under Basel I to, on a conservative esti-mate, over 200,000 under Basel II. To determine the regulatory capital ratio of this bank, the number of calculations has risen from single fig-ures to over 200 million. The quant[itative analyst] and the computer have displaced the clerk and the envelope.”

TABLE 10.3

Risk Weights for Different Categories of Credit Risk1

(Minimum, Median, Maximum)

Mortgages Corporates Institutions Other retailAutonomous 5% - 20% - 53% 32% - 59% - 76% n/a n/aBarclays 7% - 15% - 49% 33% - 55% - 89% n/a n/aBBVA 8% - 15% - 23% 37% - 52% - 78% 4% - 16% - 27% 14% - 33% - 48%BNP 6% - 13% - 25% 27% - 54% - 75% n/a 10% - 38% - 156%KBW 6% - 18% - 53% 26% - 55% - 158% 6% - 19% -34% 7% - 36% - 64%

Average 6.4% - 16.2% - 41% 31% - 55% - 95% 5% - 18% - 31% 10% - 36% - 89%

Sources: Analyst reports based on Pillar 3 disclosure; company data; and analysts’ estimates.1Institution and coverage: Autonomous = 22 European banks, two Canadian and two Australian banks (corpo-rate loans and mortgages only); Barclays = 21 European banks (corporate loans and mortgages only), full set of data for 2009 (used), as 2010 is partial: BBVA = 12 European banks; BNP = 22 European banks covered (2010 data— median); KBW = 27 European banks.

10 Under Basel I, credit risk is subject to a very broad treatment (that is, 0 percent for sovereign exposures in Organisation for Economic Co- operation and Development (OECD) member countries, 20 percent for banks, 50 percent for mortgages, and 100 percent for corporates). See also Appendix Table 10.1.1.



uniformly applied across jurisdictions (BCBS 2016a and 2019). The FRTB introduces changes in capital require-ments as a result of changes in the calculation of RWAs in Basel 2.5. According to several quantitative impact studies, market- risk capital requirements for trading book exposures have significantly increased under the provisions of Basel 2.5 and the subsequent FRTB. The evidence shows that RWAs for Basel 2.5 and III, before any mitigation actions are taken (except for one bank, net of mitigation), would increase by 36 percent on average for a sample of 14 European banks and one Asian bank (IMF 2012) (Figure 10.13).

The inconsistent implementation of Basel 2.5 across Eu-rope and the United States may have resulted in uneven risk weights. When European banks implemented Basel 2.5 as part of CRD III (by the end of 2011), the revision of the capital treatment of the trading book was still under consul-tation in the United States, which did not implement Basel 2.5 until the second half of 2012. Both regions share com-mon elements in the implementation of Basel 2.5 (for example, the introduction of stressed value at risk, incremen-tal risk charge, and the comprehensive risk measure as the new modeling components); however, some factors could cause RWAs to diverge between Europe and the United States for comparable market risk. These include:

• differences in definitions (for example, the scope of securitization), regulatory intensity of the supervi-sory model approval process;

• the imposition of surcharges (for example, correla-tion trading portfolio is applied a 15 percent of

help some Canadian banks reduce their risk weight to less than 10 percent. Global banks with a wide geo-graphic footprint tend to use a combination of ap-proaches (Basel II SA, FIRB, AIRB, and sometimes Basel I) for the relevant portions of their books, de-pending on the location of the loans (or assets).

• Corporate loans display a high degree of variation in the averages for the three regions. Europe and Asia (50 percent and 65 percent, respectively) are consider-ably below North America (85 percent), as would be expected for Basel I banks. Given the higher adminis-trative costs of lending to firms (especially small and medium- sized enterprises), there is a significant im-plicit incentive to increase high- volume standardized lending (consumer retail lending and home mortgages) and commercial real estate lending. Loans to banks show relatively similar average risk weights across the world, with moderate dispersion in all three regions.

Market risk: limited convergence

The Basel Committee also overhauled the market risk regu-latory capital framework, but global convergence remains limited. In January 2016, after a five- year consultation pro-cess, the Basel Committee completed the Fundamental Re-view of the Trading Book (FRTB) as the new framework for market risk to (1) address the shortcomings of initial revision of market risk capital framework (“Basel 2.5”), and (2) de-sign a minimum capital standard for market risk to be more

0

60

10

40

50

30

20

1. Risk-Weighted Assets: Residential Mortgages

0

120

20

80

100

60

40

2. Risk-Weighted Assets: Corporate Lending

21

6

27

7 6

50

77

5869

26

100

27

Asia Europe North America Asia Europe North America

0

30

35

5

20

25

15

10

3. Credit Risk Assets: Interbank Lending

0

120

20

80

100

60

40

4. Credit Risk Assets: Retail Lending

Asia Europe North America Asia Europe North America

24

9

31

10

27

6

51

31

74

4

100

17

Sources: Individual bank reports (Pillar 3 disclosure); and authors’ estimates.

Figure 10.12 Minimum, Maximum, and Average Risk Weights by Region for Different Categories of Credit Risk, as of December 2010 (Percent)



Basel II (current) Basel 2.5 Basel III

0

1,200

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Euro

pe

Asia

Pac

ific

150

300

450

600

750

900

1,050

Sources: Individual banks’ reports; and authors’ estimates.Note: When this analysis was completed, the final legislation governing the review of the trading book was not completed in the United States, so US banks were excluded from this chart. RWAs = risk-weighted assets.

Figure 10.13 Basel 2.5 and Basel III Impact on RWAs of Select Banks, as of December 2010 (In billions of US dollars)

ferences are largely explained by institutional, accounting, regulatory, and bank- specific factors, but the importance of unexplained factors is still quite material. The Bank of En-gland (2011a) argues that “evidence from the recent crisis suggests that the observed variations in RWAs might not en-tirely reflect genuine difference in risk taking.” The UK Fi-nancial Services Authority (2010a, 2010b) reviewed the RWA practices of UK banks in its “hypothetical portfolio exercise” (Box 10.1). In their Pillar 3 disclosure, banks tended to refer to “model changes,” “data cleansing,” “RWA optimization,” “parameter update,” or other techniques to explain a decline in RWAs.11 However, detailed disclosure is very limited. Based on publicly available data alone, it is nearly impossible to assess the extent to which variations stem from genuine changes in banks’ asset mix and risk ap-petite or from a less palatable shift in risk measurement.

RWA practices have been particularly criticized in the context of sovereign risk and covered bond exposures. Some of the key concerns regarding the latter are discussed in Box 10.2. A comprehensive discussion of sovereign risk was included in Chapter 2 of the April 2012 Global Financial Stability Report (IMF 2012). The Basel Committee has also reviewed the RWA methodology, which was included in the finalization of the Basel Accord (BCBS 2017a), and the re-view of sovereign risk (Hannoun 2011), which resulted in the publication of a discussion paper (BCBS 2017b).

standard charges surcharge to comprehensive risk measure in the US proposal, versus an 8 percent of standard charges floor in Europe); and

• the inconsistent use of external ratings (that is, the capital assessment of covered debt and securitization exposures under the standardized approach in Eu-rope is based on external credit ratings if available), whereas alternatives to ratings must be used in the United States ( Dodd- Frank Act, Section 939A).

Business models explain the remaining differences in RWAs. Banks biased toward credit trading products will at-tract higher risk weights (owing to the higher incremental risk charge) than banks geared toward flow foreign ex-change, rate products, equities, or advisory services (Figures 10.14 and 10.15). Similarly, banks with a higher proportion of secured financing will carry lower-risk weights than banks heavily exposed to securitization transactions or to princi-pal/proprietary trading. Given that the risk weights for over- the- counter derivatives have resulted in a steep rise in overall risk weights, banks also have an incentive to move toward standardized derivatives that are cleared by central counter-parties or conduct secured over- the- counter transactions.

Particular Attention on Certain RWA Differences

There is considerable scope for subjectivity and interpreta-tion in determining risk weights, which might explain their persistent variability within the same country and across na-tional boundaries. Most banks rely on a combination of ap-proaches to calculate RWAs, which inevitably entail complexity and opacity. However, cross- border comparisons may be of limited value, especially if banks have very differ-ent business profiles. Comparisons are more meaningful when targeted at banks within a relatively homogeneous re-gion or with a similar business mix.

Understanding why there are material differences and whether they are legitimate is a difficult exercise. Some dif-

11 Bruno, Nocera, and Resti (2015) find significant differences in banks’ average risk weights, both over time and across countries, based on a sample of 50 large European banks between 2008 and 2012. They at-tribute these differences to several factors, some of which reflect the ac-tual risk content of bank’s assets due to different business model and asset mix, while others might have been attributable to “RWA tweak-ing” and supervisory segmentations (for instance, the adoption of IRB approaches is [as expected] a powerful driver of RWAs, and IRB adop-tion is more widespread in countries where supervisory capture is po-tentially stronger, owing to a banking industry that is both larger [compared to GDP] and concentrated).



Trading Book Loan Book Derivatives Book

0%

90%80%

100%

10%20%30%40%50%60%70%

Asia

Pac

ific

Euro

peEu

rope

Euro

peAs

ia P

acifi

cEu

rope

Japa

nAs

ia P

acifi

cEu

rope

Euro

peEu

rope

N. A

mer

ica

Euro

peN.

Am

eric

aEu

rope

Asia

Pac

ific

N. A

mer

ica

Asia

Pac

ific

Euro

peEu

rope

Euro

peEu

rope

Euro

peEu

rope

Euro

peEu

rope

Euro

peEu

rope

Asia

Pac

ific

Asia

Pac

ific

Asia

Pac

ific

Asia

Pac

ific

Asia

Pac

ific

Asia

Pac

ific

Asia

Pac

ific

Euro

peAs

ia P

acifi

c

1. IFRS Banks: Asset Mix of Global Wholesale Division

0%

90%80%

100%

10%20%30%40%50%60%70%

N. A

mer

ica

N. A

mer

ica

N. A

mer

ica

N. A

mer

ica

N. A

mer

ica

Asia

Pac

ific

N. A

mer

ica

N. A

mer

ica

Asia

Pac

ific

N. A

mer

ica

Asia

Pac

ific

2. Non-IFRS Banks: Asset Mix of Global Wholesale Division

Sources: Bankscope; SNL; and authors’ estimates (as of June 2011).Note: IFRS = International Financial Reporting Standards.

Figure 10.14 Breakdown of Wholesale Assets: IFRS and Non- IFRS Banks, as of June 2011 (In percent)

Sources: Banks’ Pillar III reports; and authors’ estimates.Note: FX = foreign exchange; VaR = value at risk.

Figure 10.15 Value at Risk for Market Risk under Basel II, as of June 2011

VaR Interest Rates VaR FX VaR Equities VaR Commodities VaR Other

0

9080

100

10203040506070

Asia

Pac

ific

Euro

peEu

rope

Euro

peAs

ia P

acifi

cAs

ia P

acifi

cEu

rope

Asia

Pac

ific

Asia

Pac

ific

Asia

Pac

ific

Euro

peAs

ia P

acifi

cN.

Am

eric

aAs

ia P

acifi

cN.

Am

eric

aEu

rope

Euro

peEu

rope

N. A

mer

ica

Euro

peN.

Am

eric

aN.

Am

eric

aEu

rope

Euro

peEu

rope

Asia

Pac

ific

Asia

Pac

ific

Euro

peN.

Am

eric

aAs

ia P

acifi

cAs

ia P

acifi

cEu

rope

N. A

mer

ica

Asia

Pac

ific

Asia

Pac

ific

Asia

Pac

ific

Euro

peEu

rope

Euro

peAs

ia P

acifi

cAs

ia P

acifi

c

Market Risk VaR Distribution in Percent (December 2010)



Box 10.1. UK Financial Services Authority Survey of RWA Practices

To evaluate the objectivity of foundation internal- ratings- based or advanced internal- ratings- based methodology in estimating risk- weighted assets, the UK Financial Services Authority (FSA) conducted a benchmarking exercise in 2007 and 2009, called the Hypothetical Portfolio Exercise. The exercise involved 13 UK banks, and covered 50 sovereign issuers, 100 banks, and 200 large companies. It compared participants’ internal models for estimating the probability of defaults (PDs) against historical annual default rates utilized by Standard & Poor’s (1981–2008). Banks only reported their PDs for counterparties to which they had exposure. The overlap in exposures among partici-pants was limited, but the FSA identified a group (the “ jointly rated sample”) of sovereigns, banks, and corporates which all participating firms had to rate, to facilitate comparisons.

The FSA survey revealed a large dispersion in estimated PDs, suggesting that banks had very different views on the same underlying risk. The highest mean PD was three to six times larger than the lowest mean PD. Small differences could be attributed to point- in- time versus through- the- cycle features. The biggest ranges were observed on corporates, and banks to a lesser extent. Variations of PDs on sovereign exposures were much smaller.

Source: FSA (2010).

Figure 10.1.1 Variations in Estimated Probabilities of Default on Common

Hypothetical Portfolios

Interquartile rangeAverage

Maximum-minimum range

0.00

0.05

0.10

0.15

0.20

Sovereign Bank Corporate

Estimated mean probability of default (percent)

There are several “red flags” relative to risk measurement and the calculation of RWAs, which may need to be investi-gated if they occur as part of a bank’s regular reporting:

• RWAs experience large swings over time without material changes in the bank’s business mix.

• A strong Basel III risk- weighted capital ratio coin-cides with a weak leverage ratio (based on what can be inferred from disclosed information during the observation period).

• Downgrading from a more sophisticated risk assess-ment approach (for example, AIRB) to take advan-tage of simplifications in less sophisticated approaches (for example, uniform 2.5-year maturity assumption under the FIRB or the application of a zero percent risk weight for sovereign risk under the standardized approach).

• “ Cherry- picking” the most favorable methodology for each type of asset type and exposure with a view to optimizing capital intensity rather than respond-ing to the coexistence of various operating environ-ments in different jurisdictions.

• Credit RWAs decline during economic downturns (if the bank reports under the IRB approach).12

• RWAs are lower than other local banks or banks with the same business model, suggesting that banks support comparable or identical risks with very dif-ferent levels of capital.

• Significantly lower risk weights are not verifiable and/or justified based on “methodology changes” without further explanation.

5. WHAT CAN BE DONE TO RESTORE CONFIDENCE IN RWAS?

Objectives of RWA Reforms

This chapter reviewed the various causes for variations of RWAs across banks. While some differences are plausible, publicly available data suggests that there are some areas

12 The directional effect of economic downturns on market risk is not as straightforward and, thus, cannot be generalized.



Box 10.2. Are There Some Anomalies in the Treatment of Covered Bonds?

Basel II/III treat covered bonds in the same manner as bank unsecured debt for risk- weighting purposes (see first row of Table 10.2.1). How-ever, for European banks using the revised standardized approach, certain “Capital Requirement Directive (CRD)-compliant” covered bonds issued by European Union banks can be assigned a risk weight as low as 10 percent. The bonds must meet the requirements of Article 52(4) of Directive 85/611/EEC (UCITS 52(4)), and be secured by eligible assets prescribed under the European CRD. For instance, a loss given default (LGD) of between 11.25 percent and 12.5 percent can be applied to eligible covered bonds (meeting certain criteria), whereas under Basel II, the LGD is 45 percent, similar to senior unsecured bank debt.

There are important implications in using the standardized versus internal- ratings- based approach. Table 10.2.1 presents the standard-ized approach, where weights depend on the credit rating of the issuing institution, so that a CRD- compliant covered bond issued by a bank rated single- A is assigned a 20 percent weight, versus 50 percent on its unsecured debt. The table assumes that the relevant jurisdic-tion has mandated the use of the credit assessment- based method (“Option 2”) where the risk weight of an issuer’s senior unsecured debt depends on the issuer’s external credit rating. The central government risk weight- based method (“Option 1”) bases the risk weight of an issuer’s senior unsecured debt on the external credit rating of the central government of the jurisdiction in which the issuer is incorpo-rated. Under the IRB approach, risk weights for covered bonds are derived from the probability of default of the issuer or sponsor bank (which is itself linked to its senior unsecured rating.)

The favorable treatment of covered bonds is a European specificity. The preferential treatment of covered bonds is an area of conten-tion. Concerns have emerged on the lack of differentiation in LGD and risk weight between covered bonds backed by assets deemed risky (such as shipping, commercial property, and so on) and those backed by more stable and higher quality assets (such as prime residential mortgages and public sector collateral). Another weakness stems from the uneven treatment of covered bonds and securitizations, with the former benefiting from a much more favorable treatment, thus encouraging banks to package reference portfolios for asset- backed securities into covered bonds. An additional concern is the link between covered bond risk weights and sovereign ratings, allowing cov-ered bonds to continue to receive a preferential risk weight even if the issuing bank is downgraded (as long as the sovereign rating of the country of issues remains above a certain threshold.)

Within Europe, not all countries follow CRD rules, and some have adopted a stricter framework, increasing the differences in risk- weight approaches across jurisdictions. At a minimum, and until a full review of covered bond risk- weight practices is conducted, banks should be required to disclose their covered bond exposures and to indicate whether they have benefited from any preferential treatment under CRD provisions.

TABLE 10.2.1

Basel II Risk Weights for Senior Unsecured Debt and Covered BondsOriginating Institution’s Credit Rating

AAA/AA A BBB BB B

Senior unsecured debt and non- CRD- compliant covered bonds

20% 50% 50% 100% 150%

CRD- compliant covered bonds 10% 20% 20% 50% 100%

Source: European Commission.Note: CRD = Capital Requirement Directive.

easily manipulated. Some of these considerations are already reflected in the finalized version of the Basel III framework (BCBS 2017a), but some areas remain contested, such as the treatment of sovereign exposures, and the country- specific phased- in implementation of a stricter framework to allow banks and supervisors time to build in sufficient capacity to implement and validate internal models.

Overall, the implementation of reforms should satisfy the following criteria:

• Balance comparability versus flexibility: Regulators will need to trade off greater comparability across banks and jurisdictions (but possibly ignoring spe-cial situations for certain asset classes or in certain countries) against retaining sufficient flexibility at the bank/country level in calibrating RWAs.

• Level of implementation: For global banks with oper-ations in multiple jurisdictions, a high degree of

where the variations seem less justified. Some RWA differences are structural, reflecting banks’ different business models and risk profiles, but also account for different views on risks among regulators. In addition, some variation of RWAs for similar risks may even be financial stability- enhancing if they encourage the coexistence of banks’ differ-ent business choices, models, and risk appetites. Conversely, uniform risk measurement might lead to herd behavior and be detrimental to some asset classes if all banks reduced their exposures. The objective of reforming RWAs is not to seek a full harmonization but to ensure greater consistency of methodologies, comparability of risk measures, and trans-parency in disclosure to supervisors and markets.

This section provides several considerations for enhanc-ing the reliability of a risk- based framework for banking regulations through a robust alignment of capital allocation and asset pricing with underlying risks, which cannot be



Strengthening disclosure complements effective regula-tory and supervisory measures. Limited and inconsistent dis-closure of RWA assumptions and methodologies makes it difficult to compare banks and determine whether different risk assessments are justified. Providing more granular infor-mation on a frequent basis would foster market discipline, and in the short term, could help banks reduce investor un-certainty (Table 10.4). It also involves the choice of how mar-ket consistency can be achieved in the calibration of RWAs. The two extreme (Figure 10.17) views are that (1) banks and credit rating agencies (CRAs) are best placed to measure risks, or (2) regulators should determine the floors for each asset, regardless of banks’ internal risk assessments. These drive policy options, which range from (1) a full, unfettered modeling within banks; to (2) a prescriptive regulatory ap-proach relying on prudential floors.

While a more standardized approach may carry the benefit of simplicity, it may not necessarily capture properly the cost of risk (for example, on sovereign risk) and a “ one- size- fits- all” may not lead to greater financial stability. Regulators already have tools (including Pillar II), which can influence the attrac-tiveness of approaches (for example, imposing additional floors on the IRB approach could reduce its desirability versus revamping a standardized approach). Alternatively, banks could use their internal models to calculate RWAs, but regula-tors and supervisors would address possible deficiencies and contain excessive bank discretion, where necessary and under clear guidelines.14

The overall capital assessment of banks would ideally rely on multiple capital metrics. Risk- based measures should be used in tandem with less risk- sensitive capital measures to capture the riskiness of exposures while limiting excessive leverage. The new Basel III leverage ratio serves as an impor-tant backstop to risk- based ratios that rely extensively on banks’ internal models. Linking capital ratios to the macro-prudential framework would be useful in addressing exuber-ant or constrained asset classes.

Overall, capital adequacy is but one, albeit key, part of a holistic approach to assessing banks’ financial strength. Capital requirements should not be viewed as a “ one- size- fits- all” benchmark, but, rather, should be tailored to match the level of credit, market, and operational risk each bank is taking. Other risks should also be considered, such as un-derwriting policies, asset classification and hedging, provi-sioning measures, and concentration risks. An effective RWA reform will require a combination of measures based on Pillars I through III, as well as on enhanced internal risk management (Table 10.4).

international cooperation is essential, and the Basel framework has been designed with “internationally active banks” in mind. For local banks with tradi-tional lending activities, reforms should be propor-tionate and more nationally or regionally focused, and international guidance could be aimed at foster-ing greater convergence in methodologies.

• Prioritization on the most pressing deficiencies: The Ba-sel Committee has reviewed the measurement of RWAs in both the banking and trading books; how-ever, the treatment of sovereign risk remains unad-dressed as an important element in restoring the credibility of the Basel II standardized and IRB ap-proaches. On market risk, the FRTB prompted some changes, but implementation varies across regions and can result in significant inconsistencies.

• Impact on the stringency of capital ratio: Until now, many regulators have compensated for what they re-gard as insufficient risk recognition in RWAs through higher capital ratios. However, if risk is to be better measured going forward, capital require-ments may need to be reviewed so that a tightening in the denominator is not amplified by the implicit “risk premium” in current capital ratios.

• Consistency with the broader agenda of postcrisis regu-latory reforms: The reform of RWAs should not con-flict with other regulatory changes, for instance liquidity rules (especially for sovereign risks) and macroprudential rules.13 More dynamic risk weights can help dampen asset bubbles but also increase the procyclicality of credit growth, and, thus, can im-pact the effectiveness of the countercyclical capital buffer (for example, by imposing prudential floors to prevent erosion through IRB models in good times or raising specific risk weights in the face of mount-ing risks or uncertainty).

Multi- pronged Approach to Reforming RWAs

The effective reform of RWAs would need to rely on a com-bination of measures. These include (1) regulatory changes to the existing capital framework, (2) more intrusive super-visory intervention, (3) additional disclosure by banks to fos-ter greater market discipline, and (4) more robust risk management and strategy within banks (Figure 10.16).

Enhanced supervision is essential for this effort. Pillar II offers supervisors extended powers to scrutinize risk weights. However, efficient oversight requires commensurate supervi-sory capacity at the country level. In the absence of skilled and sufficient staff, as well as a clear mandate and structure, a logi-cal safeguard would be to either (1) require IRB banks to ap-ply more conservative haircuts on assets, or (2) restrict their use of internal models in favor of standardized approaches.

13 See also Bank of England 2011b.

14 Banks need to rely more on an economic rather than regulatory ap-proach for measuring and managing risk. The Boston Consulting Group (2011) finds that banks are excessively focused on ensuring com-pliance with capital requirements rather than managing risk costs and creating economic value. Banks report risk metrics, for example, but have not integrated them into key business processes or used them to influence critical business decisions.



Revisedregulations

Enhancedsupervision

Improvedmarket

discipline

Enhanced riskmanagement

Governance& financial strength

Risk management& risk profile

Capital allocation& asset pricing

RWA & assetmix

Source: Authors.Note: RWA = risk-weighted asset.

Figure 10.16 Reforming RWAs Has to Rely on a Combination of Measures

Back toBasel I

LeverageRatio only

Basel IIStandardizedApproach only

Basel II IRBmaintained withtransition floors

Basel IIA-IRB only

Introductionof floors &

review of RW

Source: Authors.Note: IRB = internal- ratings- based.

Figure 10.17 Possible Options to Reform the Existing RWA Framework

TABLE 10.4

Some Policy Options to Revisit the RWA FrameworkPolicy Options Comments

Pillar 1—Reforming Quantitative MeasuresIntroduce or increase

minimum prudential floors

Floors can be introduced at different levels, within the IRB formula, on RWAs or on capital ratios. Higher floors on the IRB approach could reduce its attractiveness (in terms of lower risk weights and lower capital requirements) versus the standardized approach, and encourage banks to be more prudent and conservative.

• PD, should be estimated “ through- the- cycle” rather than as “ point- in- time” measures, accounting for crisis periods and tail risks.

• On LGD, some countries may consider topping up the 10 percent floor in Basel II.• On M, the catch- all 2.5-year average maturity under FIRB does not reflect the long duration of the

mortgage portfolio of most retail banks and should be discouraged.• On RW, introducing minimum floors (including temporary ones) would be appropriate for asset classes that

experience excessive growth, or for assets deemed riskier (for instance, resecuritization or commercial real estate). Conversely, regulators may also continue to give preferential treatment to encourage certain types of lending (lower RWs for loans to SMEs).

Allow for time- varying or variable credit risk weights

In conjunction with macroprudential policies, rising (declining) RWs help address excessive (insufficient) credit growth. Higher RWs could also target emerging risks in specific exposures and distinguish flows from stocks to directly influence lending. For IRB banks, a multiplicative scalar could be applied to internal model outputs.

Allow for a variable scaling factor

RWs would stay constant, but Basel II’s 1.06 scaling factor could be adjusted upward or downward to divert or attract capital to certain asset classes.

Return to standardized approaches and abandon the IRB approach

Advantages • Simple approach, more transparent, and easy to monitor.• Some risk sensitivity (that is, more than Basel I).Disadvantages• Inconsistent with regulatory incentive to encourage banks to develop advanced risk management through

internal models.• Higher reliance on CRAs in assessing credit risk (which also creates comparability challenges in countries

where the use of CRAs for capital assessments is no longer permitted, for instance, in the United States).• Limited differentiation for some asset classes (for example, all mortgages are equally weighted).• Requires more capital than under IRB approaches (by construction), especially for corporate and mortgage

exposure.



TABLE 10.4 (continued)

Some Policy Options to Revisit the RWA FrameworkPolicy Options Comments

Exclusive reliance on a leverage ratio

Advantages• Simple to compute and more objective.• Safeguard against model risk and measurement errors eroding the effectiveness of a risk- based capital

measure.• Prevents banks from arbitraging different assessment approaches for credit, market, and operational

risks.• Imposes an implicit RW floor (depending on the risk- based capital ratio), which makes it less attractive

for banks to accumulate “low or zero risk” assets.Disadvantages• Does not count for different asset portfolios across banks.• Does not differentiate between risky and safer assets.• Does not help optimize the level of capital based on reliable risk measures underpinning the capital

assessment.Exclusive reliance on

AIRB and remove floor constraints

Advantage Efficient use of capital since banks would use their own internal models, unfettered by floors or caps.DisadvantageBanks have incentive to reduce risk weights to maximize return on capital without consideration of

sufficient buffers and level of sufficient conservatism to ensure solvency during times of stress.

Pillar 2—Enhancing SupervisionGreater use of Pillar II at

individual bank or system- wide level

The capital review under Pillar 2 allows supervisors to adjust a bank’s RWs if they (1) are considered too low (based on the bank’s business model), or (2) significantly deviate from peers’ best practices. A correction of RWs might also be warranted as an add- on buffer to mitigate modeling risk. Supervisors should be able to build in additional buffers, by requiring higher RWs (and more capital) on assets they deem riskier.

Enhanced monitoring and remedial powers

Supervisors need to continuously assess (not just in the prevalidation phase) the robustness of banks’ models and suggest remedial actions if necessary.

Harmonize implementation of existing rules

Supervisors need to develop appropriate and transparent standards for the calibration of risk weights, which would also facilitate the international harmonization of supervisory practices.

Supervisory peer review and cross- country monitoring

The Basel Committee created the Regulatory Consistency Assessment Programme and sponsors regular Peer Reviews (BCBS 2012) among its membership. The horizontal reviews during the transition to the finalized Basel III framework (1) help assess the level of variation across supervisory practices, and (2) ensure a consistent and comparable implementation of the Basel rules. In addition, national supervisors could participate in joint validation teams of global banks with important local presence to verify the methodologies used at individual (international) banks.

Pillar 3—Fostering Greater Market DisciplineGreater market disclosure

to enhance market confidence

The disclosure of more granular and frequent (quarterly) information on the composition of RWAs, ideally broken down by geography, improves the transparency of banks’ risk assessment. Such disclosure could also include explanations of internal risk models and any material changes that occur over time. The latest edition of the European Banking Authority’s Transparency Exercise (EBA 2018) illustrates how regulators could be involved in this process by requiring a minimum degree of disclosure on an annual basis.

Improved communication The dialogue between banks, investors, and industry analysts could be improved, especially on changes in models, exemptions, or material reduction in RWA levels.

Enhanced role for bank auditors

Auditors are both independent third parties and have access to confidential bank information. Their reports could also include assessments of risk taking, risk appetite, and risk management.

Improving Banks’ Internal Risk ManagementAdopt a more “economic”

approach for measuring capital and managing risk

A closer alignment of economic and regulatory capital models (1) helps integrate more closely business decisions with cost of risk and of capital, (2) fosters a more dynamic measurement of risks than provided by ratios, and (3) would be more efficient than a purely regulatory approach for RWA calculation, capital allocation, and asset pricing.

Develop more forward- looking risk- measurement metrics

Banks would benefit from developing more dynamic, forward- looking models reflecting likely changes in their risk profiles and portfolios.

Develop models that better capture tail risk

Banks need to factor in extreme stress scenarios and have contingency plans to deal with them.

Abandon IRB and move back to standardized approaches

Banks that report under IRB may voluntarily revert back to standardized approaches to demonstrate their commitment to transparency, simplicity, and objectivity if lower costs of capital due to enhanced investor confidence offsets higher resulting capital requirements.

Source: Authors.Note: AIRB = advanced internal- ratings- based; CRA = credit rating agency; FIRB = foundation internal- ratings- based; IRB = internal- ratings- based; LGD = loss given default; M = effective maturity; PD = probability of default; RW = risk weight; RWA = risk- weighted asset; SMEs = small and medium- sized enterprises.



banks’ business models, risk profiles, and riskmanagement approaches but are also due to different regulatory frame-works and supervisory regimes. Since full harmonization and convergence of RWA practices may not be achievable, the focus of RWA reforms should be squarely placed on (1) improved transparency and understanding of outputs and (2) common guidance on RWA methodologies. A multi-pronged approach would be required, combining regulatory changes, enhanced supervision, and greater market disclo-sure to maximize the benefits of risk- based capital ratios.

6. CONCLUSIONPerceived differences in RWAs within and across countries have diminished the reliability, consistency, and comparabil-ity of RWAs and capital ratios, and if not addressed, could degrade the credibility of the regulatory framework. This chapter explored the variation of RWAs between banks within and across regions and provided several consider-ations for the effective implementation of the finalized Basel III framework.

The findings suggest that several factors drive differences in RWAs. Some of these differences are not only the result of


Appendix 10.1.Evolution of the Regulatory

Capital Framework

FROM BASEL I TO BASEL II: A NEW WAY TO CALCULATE RISK-WEIGHTED ASSETSIn 1988, the Basel Committee introduced the Basel I framework, which originally dealt only with credit risk; market risk was added later. For credit risk, assets were assigned a level of capital (“risk weights”) based on the nature of the assets, ranging from zero percent for the assets deemed “safest” to 100 percent for the riskiest assets. Between 2004 and 2009, Basel II revised the way risk- weighted assets (RWAs) were computed and broadened the coverage of risks by including operational risk.15 Basel II assigns risk weights based on the quality of assets, as measured either by external ratings provided by external credit rating agencies or by internal ratings calculated by banks, based on their own internal models.

After the global financial crisis, the Basel Committee revised Basel II to what evolved into the current Basel III framework (with interim amendments captured in Basel 2.5), which was finalized in December 2017. Basel III mainly strengthens the numerator of the capital ratio through a stricter definition and composition of the capital while the changes to the denominator (that is, RWAs) are more limited. It raises the minimum capital requirements (to 7 percent, including the capital conservation buffer) and introduces both (1) a countercyclical capital buffer (variable, of up to 250 basis points), and (2) an international leverage ratio requirement. In addition, global systemically important banks are subject to an additional capital requirement of between 1 and 2.5 percent of RWAs (BCBS 2013c).

Basel III paves the way to greater consistency of RWA related to market risk in Basel 2.5 and the Fundamental Review of the Trading Book, which introduced new requirements for the assessment of market risk. The three most important changes were: (1) the incremental risk charge, a charge designed to capture migration and default risk for securities within the trading book (for example, bonds, credit derivatives, and leveraged loans; (2) the stressed value- at- risk capital charge (which comes in addition to the current requirements), designed to account for volatile market conditions; and (3) the credit valuation adjustment charge, aimed at capturing counterparty credit risk. These amendments to the existing framework were finalized in 2016 and reduced capital arbi-trage opportunities between the trading and banking books, in particular for securitization exposures (BCBS 2016a, 2019).

The Basel framework has become increasingly granular over time, especially for credit risk. Under Basel I, credit risk was categorized into five risk-weight buckets (0, 10, 20, 50, and 100 percent) for four different asset categories: claims on sover-eigns, banks, residential mortgage loans, and firms. The 100 percent risk bucket was considered the “normal risk” bucket, but preferential weights are given to claims on governments or banks in Organisation for Economic Co- operation and Develop-ment member countries. Basel II transitioned the framework from the one- size- fits- all approach to a tailor- made approach to bank capital. The concept was to offer an “evolutionary approach” to calculate regulatory capital. Therefore, banks that could meet a series of quantitative and qualitative criteria would be allowed to use more risk- sensitive, internal- model- based method-ologies. Simpler standardized methodologies would also be offered in a menu of options, with built- in incentives for banks to improve their riskmanagement practices and risk measurement and therefore qualify for the more advanced approaches. Un-der Basel II (and Basel III), credit risk is calculated based on three different approaches of varying degrees of sophistication: the standardized approach as well as both internal- ratings- based (IRB) approaches (foundation and advanced) for corporate expo-sures, including the slotting criteria approach for specialized lending under the foundation IRB approach.

Standardized Approach

Under the standardized approach, Basel II risk weights can be significantly higher than 100 percent (for exposures rated “ B-” and lower), which was general risk weight under Basel I, but they can also be much lower, especially for investment- grade- rated exposures (see Appendix Table 10.1.1). Low credit ratings increase risk weights for sovereign counterparties to as high as 150 percent (from 0 percent under Basel I), banks and securities firms (from 20 percent), and nonfinancial corporate

15 For operational risk, there are also three different approaches: (1) the basic indicator approach, (2) the standardized approach, and (3) the advanced mea-surement approach. The last approach was abandoned in the finalized version of the Basel III framework.



counterparties (from 100 percent). However, risk weights are as low as 30 percent for residential mortgages with loan- to- value ratios up to 50 (80) percent (if the repayment materially depends [does not depend] on cash flows generated by property), and for consumer retail lending from 100 to 75 percent. Moreover, good ratings can reduce risk weights for exposures to nonfinan-cial firms and commercial real estate to as low as 20 percent (from 100 percent under Basel I).


Illustration of Risk Weights from Basel I to Basel IIBasel I Basel II/III Standardized Approach

Claims on sovereigns (and central banks) and public sector entities1

• OECD: 0%• Non- OECD: 100%

Exposures to own sovereign in domestic currency (0%), subject to national discretion

• AAA to AA−: 0%• A+ to A−: 20%• BBB+ to BBB−: 50%• BB+ to B−: 100%2

• Below B−: 150%• Unrated: 100%National discretion for exposures to own sovereign in domestic currency: 0%

Claims on multilateral development banks or other international financial institutions

20% • Multilateral development banks3 • IMF, BIS, ECB, and EC: 0%• Otherwise based on Option 2 (based on internal ratings) for claims on banks

(see below)

Claims on banks Short term (<1 year)• OECD: 20%• Non- OECD: 20%Long term (>1 year)• OECD: 20%• Non- OECD: 100%

Option 1 (based on sovereign rating minus one RW category)• AAA to AA−: 20%• A+ to A−: 50%• BBB+ to BBB−: 75%• BB+ to B−: 100%• Below B−: 150%• Unrated: 100%Option 2 (based on internal bank rating)• Short term (<3 months)

AAA to AA−: 20% A+ to A−: 20% BBB+ to BBB−: 20% BB+ to B−: 50% Below B−: 150% Unrated: 20%

• Long term (>3 months) (“Base Risk Weight”) AAA to AA−: 20% A+ to A−: 30% BBB+ to BBB−: 50% BB+ to B−: 100% BB+ to B−: 100% Below B−: 150% Unrated: 50%

Mortgages 50% for residential properties occupied (or rented) by borrower and secured by first charge on property

Residential mortgage: Repayment not materially dependent on cash flows generated by property

LTV ≤ 50%: 20%50% < LTV ≤ 60%: 25%60% < LTV ≤ 80%: 30%80% < LTV ≤ 90%: 40%90% < LTV ≤ 100%: 50%LTV > 100%: 70%Residential mortgage: Repayment materially dependent on cash flows generated

by property• LTV ≤ 50%: 30%• 50% < LTV ≤ 60%: 35%• 60% < LTV ≤ 80%: 45%• 80% < LTV ≤ 90%: 60%• 90% < LTV ≤ 100%: 75%• LTV > 100%: 105%Residential commercial mortgage: Repayment not materially dependent on cash

flows generated by property• LTV ≤ 60%: minimum (60%, RW of counterparty)• LTV > 60%: RW of counterpartyCommercial mortgage: Repayment materially dependent on cash flows generated

by property• LTV ≤ 60%: 70%• 60% < LTV ≤ 80%: 90%• LTV > 80%: 110%




Illustration of Risk Weights from Basel I to Basel IIBasel I Basel II/III Standardized Approach

Claims on firms (including insurance)

100% National discretion to weigh all corporate claims at 100%, or to use ratings based on:

• AAA to AA−: 20%• A+ to A−: 50%• BBB+ to BBB−: 75%• BB+ to BB−: 100%• Below BB−: 150%• Unrated: 100%For qualifying, unrated (retail) SME exposures: 85% (75%)

Source: BCBS 2017a.Note: BIS = Bank for International Settlements; EC = European Commission; ECB = European Central Bank; LTV = loan- to- value ratio; MDBs = multilateral develop-ment banks; OECD = Organisation for Economic Co- operation and Development; PSEs = public sector entities; RW = risk weight; SME = small and medium- sized enterprise.1Subject to national discretion, exposures to certain domestic PSEs may also be treated as exposures to the sovereigns in whose jurisdictions the PSEs are established.2Would be 50 percent under Option 2 (based on external rating of PSE). 3MDBs currently eligible for a 0 percent risk weight are: the World Bank Group (comprising the International Bank for Reconstruction and Development, the International Finance Corporation, the Multilateral Investment Guarantee Agency, and the International Development Association); the Asian Develop-ment Bank, the African Development Bank, the European Bank for Reconstruction and Development, the Inter- American Development Bank, the Euro-pean Investment Bank, the European Investment Fund, the Nordic Investment Bank, the Caribbean Development Bank, the Islamic Development Bank, the Council of Europe Development Bank, the International Finance Facility for Immunization, and the Asian Infrastructure Investment Bank.

Internal- Ratings- Based Approach

Basel II introduced the IRB approach (BCBS 2005b). The IRB approach is based on the concept that banks with better riskmanagement capacity (meeting a range of conditions and under explicit supervisory approval) would be allowed to use their own classifications and measurements for “key drivers” of credit. The IRB is offered in two versions: (1) the foundation IRB approach, under which banks can estimate the probability of default (with other credit risk parameters being predetermined),16 and the advanced IRB approach, under which banks are also allowed to use their own calculations of loss given default, exposure- at- default, and effective maturity. The IRB approaches stop short of allowing banks to use their own models to calcu-late the actual capital requirements. Instead, the measures of the key drivers are converted into risk weights and capital require-ments by using regulatory formulas. Different risk- weight functions are provided for each category of assets (firms, banks, sovereigns, retail, and equity), with adjustments for asset correlations and maturity.

The IRB approaches provides a strong economic incentive for banks to adopt internal riskmanagement practices for prudential purposes. The risk sensitivity embedded in the framework generally results in a lower capital intensity of credit risk than the stan-dardized approach (Appendix Figure 10.1.1). However, the Basel Committee also built in a few safeguards. In addition to the re-quired supervisory approval for using IRB approaches, Basel II introduced (1) a scaling factor of 1.06 to be applied to RWAs calculated under the IRB approach, and (2) prudential floors based on what would be the capital requirement under Basel I.

16 If banks do not meet the requirements for estimating the probability of default under the foundation IRB approach, they are required to map their inter-nal risk grades to five supervisory categories, each of which is associated with a specific risk weight (the “slotting criteria approach”). These supervisory categories (from “strong” to “weak” and “default”) represent qualitative proxies of credit grades. The risk weights associated with each supervisory cate-gory range from 70 to 250 percent.

Source: Authors.Note: CAR = capital adequacy ratio; IRB = internal-ratings-based; min = minimum.

Appendix Figure 10.1.1 Regulatory Capital— Corporate Risk Credit

Basel I (at min. CAR = 8%)Basel II (at min. CAR = 8%)Basel III (at min. CAR = 10.5%)IRB (Basel III)

0

18

AAA B–BB+BB–BBBB+BBB–BBBBBB+A A–A+AA–AAAA+

4

2

6

8

10

12

14

16

Perc

enta

ge o

f Exp

osur

eRe

quire

d to

Be

Held

as

Capi

tal

Rating Category


Appendix 10.2.Methodology and Sample

Description

The analysis in this chapter relies on published financial data of 50 systemically important banks in 19 countries.17 The sample banks are grouped by region (North America [12 banks], Europe [20], and Asia Pacific [18]), business model (investment banks [7], retail banks [12], and universal banks [31]) as well as regulatory reporting and accounting standards prevalent in their home jurisdictions. The total assets of European sample banks exceed those of their Asia Pacific and North American peers by 136 percent and 149 percent, respectively. Similarly, the European banks aggregate risk- weighted assets and tangible common equity are both higher than those of Asian Pacific (by 71 and 69 percent) and North American (by 46 and 38 percent) banks. The deviation in European total assets appears much wider than in risk- weighted assets or tangible common equity.

0

14

2

4

6

8

10

12

1. Number of Banks by Region 3. Number of Banks by Region and Credit Rating

AsiaPacific

18

Europe20

NorthAmerica

12Universal

Banks31

RetailBanks

12

InvestmentBanks

7

2. Number of Banks by Business Model

AA AA–

A+A A–

BBB+ BB

BBB

B–

0

35

5

10

15

20

25

30

4. Aggregate Total Assets (Trillions of US dollars)

AP EUR NA0

12

2

4

6

8

10

5. Aggregate Risk-Weighted Assets (Trillions of US dollars)

AP EUR NA0

1200

200

400

600

800

1000

6. Aggregate Tangible Common Equity (Billions of US dollars)

AP EUR NA

Sources: Bloomberg; individual bank reports; and authors’ estimates.Note: AP = Asia Pacific; EUR = Europe; NA = North America.

Appendix Figure 10.2.1 Global Sample of Banks, as of 2011

17 Data were obtained from individual banks’ reports, Bloomberg LP, Bankscope, SNL, and from the three main credit rating agencies, Moody’s, Standard and Poor’s, and Fitch Ratings.




Sample of Banks by Region and CountryRegion Country Bank Name

Asia Pacific Australia ANZ Banking GroupAsia Pacific Australia Commonwealth Bank of AustraliaAsia Pacific Australia Macquarie GroupAsia Pacific Australia National Australia BankAsia Pacific Australia Westpac Banking CorpAsia Pacific P.R. China Bank of ChinaAsia Pacific Hong Kong SAR Bank of East AsiaAsia Pacific Hong Kong SAR Hang Seng Bank (part of HSBC)Asia Pacific India ICICI BankAsia Pacific India State Bank of IndiaAsia Pacific Japan Mitsubishi UFJ FGAsia Pacific Japan Mizuho FGAsia Pacific Japan Nomura Holdings Inc.Asia Pacific Japan Sumitomo Mitsui FGAsia Pacific Singapore DBS Group HoldingsAsia Pacific Singapore United Overseas BankAsia Pacific South Korea Kookmin BankAsia Pacific South Korea Woori BankEurope United Kingdom BarclaysEurope United Kingdom HSBC HoldingsEurope United Kingdom Lloyds Banking GroupEurope United Kingdom Royal Bank of ScotlandEurope France BNP ParibasEurope France Crédit AgricoleEurope France Groupe BPCEEurope France Société GénéraleEurope Germany CommerzbankEurope Germany Deutsche BankEurope Italy Intesa SanpaoloEurope Italy UniCreditEurope Spain Banco Bilbao Vizcaya ArgentariaEurope Spain Banco SantanderEurope Denmark Danske BankEurope Netherlands ING BankEurope Norway Danske Bank (Norway)Europe Sweden NordeaEurope Sweden Skandinaviska Enskilda BankenEurope Switzerland Credit Suisse GroupEurope Switzerland UBSNorth America Canada Bank of MontrealNorth America Canada Bank of Nova ScotiaNorth America Canada Royal Bank of CanadaNorth America United States Bank of AmericaNorth America United States CitigroupNorth America United States Goldman SachsNorth America United States JPMorgan ChaseNorth America United States Morgan StanleyNorth America United States PNC Financial ServicesNorth America United States SunTrust BanksNorth America United States US BancorpNorth America United States Wells Fargo

Source: Authors.


Appendix 10.3.S&P’s Risk- Adjusted Capital

Framework

In 2009, Standard & Poor’s (S&P) developed its own solvency measure— the risk- adjusted- capital (RAC) ratio, to facilitate the comparability and transparency of banks’ capital adequacy across banks operating under different jurisdictions and business models. After accounting for diversification effects, the RAC ratio is broadly comparable to Core Tier 1 (which is roughly comparable to what has now evolved into the Common Equity Tier 1 under Basel III), with some country variations (Appen-dix Figure 10.3.1).

The RAC ratio is broadly consistent with the definition of capital under Basel III (and excludes hybrid instruments that would have been permissible under Basel II) and applies a stricter deduction from capital for pension deficits. However, the methodology for calculating risk- weighted assets differs from that in Basel III despite significant convergence between the two capital measures (RAC ratio is only 3 percent higher on average) (Appendix Figure 10.3.2). The RAC ratio adjusts for concen-tration and diversification of credit, market, operational risks, and insurance risks, whereas the Basel formula assumes infinite granularity of exposures. Adjustments relate to single name concentration; industry- sector diversification in the corporate portfolio; and concentration by geographies, business lines, and risk types.

While the RAC ratio fosters greater comparability, it is also constrained by the fact that (1) it covers only banks with an S&P rating; (2) the methodology is proprietary and not shared with market participants; and (3) some regulators and market participants may be uncomfortable with a higher reliance on credit rating agencies.

Sources: Bloomberg; and Standard & Poor’s.

Appendix Figure 10.3.1 Comparison of Solvency Measures: RAC Ratio, Basel II Core Tier 1, and Leverage Ratio, as of December 2010

RAC ratio after diversification Core Tier 1 capital ratio (2010) Leverage ratio (2010)

0

16

2

4

6

8

10

12

14

Aust

ralia

Neth

erla

nds

Unite

d Ki

ngdo

m

Spai

n

Hong

Kon

g SA

R

Sing

apor

e

Swed

en

Norw

ay

Cana

da

Unite

d St

ates

Fran

ce

Italy

Switz

erla

nd

Japa

n

Germ

any

Denm

ark



Government and central banks Institutions CorporateRetail Securitization Other assetsEquity in the banking book Trading book market risk Total operational risk

Basel II RWAs (end-2010) S&P RWA (end-2010)1%

5%

44%

22%

2%

6%

6%

4%10%

1%

4%

39%

21%

6%

5%

8%

7%

9%

Sources: Bloomberg; and Standard & Poor’s.Note: RWA = risk- weighted asset.

Figure 10.3.2 Comparison of RWAs: Basel II vs. Standard & Poor's, as of December 2010



REFERENCESBank of England. 2011a. “Financial Stability Report: Issue 30.”

December, Bank of England, London, UK. https://www .bankofengland.co.uk/-/media/boe/files/financial-stability -report/2011/december-2011.

———. 2011b. “Instruments of Macroprudential Policy.” Discus-sion Paper, December, Bank of England, London, UK. https://www.bankofengland.co.uk/-/media/boe/f iles/news/2011 /december/ instruments- of- macroprudential- policy- discussion -paper.

Basel Committee for Banking Supervision (BCBS). 1988. “Inter-national Convergence of Capital Measurement and Capital Standards.” BCBS Publication No. 4, October, Bank for Inter-national Settlements, Basel, Switzerland. https://www.bis.org /publ/bcbs04a.htm.

———. 1996. “Amendment to the Capital Accord to Incorporate Market Risks.” BCBS Publication No. 24, January, Bank for International Settlements, Basel, Switzerland. https://www.bis .org/publ/bcbs24.htm.

———. 1999. “Capital Requirements and Bank Behaviour: The Impact of the Basel Accord.” Working Paper No. 1, April, Bank for International Settlements, Basel, Switzerland. https://www .bis.org/publ/bcbs_wp1.htm.

———. 2005a. “Application of Basel II to Trading Activities and the Treatment of Double Default Effects.” BCBS Publication No. 116, July, Bank for International Settlements, Basel, Swit-zerland. https://www.bis.org/publ/bcbs116.htm.

———. 2005b. “An Explanatory Note on the Basel II IRB Risk Weight Functions.” July, Bank for International Settlements, Basel, Switzerland. https://www.bis.org/bcbs/irbriskweight.htm.

———. 2005c. “Basel II: International Convergence of Capital Measurement and Capital Standards, a Revised Framework.” BCBS Publication No. 118, November, Bank for International Settlements, Basel, Switzerland. http://www.bis.org/publ/bcbs118 .htm.

———. 2006. “Basel II: International Convergence of Capital Measurement and Capital Standards: A Revised Framework— Comprehensive Version.” BCBS Publication No. 128, June, Bank for International Settlements, Basel, Switzerland. https://www .bis.org/publ/bcbs128.htm.

———. 2009. “Basel II: Revised International Capital Framework.” December, Bank for International Settlements, Basel, Switzer-land. http://www.bis.org/publ/bcbsca.htm.

———. 2010. “ Overview— Enhancements to the Basel II Frame-work, including the Capital Regime for Trading Book Positions.” December, Bank for International Settlements, Basel, Switzer-land. http://www.bis.org/publ/bcbs/basel2enh0901.htm.

———. 2011a. “Final Elements of the Reforms to Raise the Qual-ity of Regulatory Capital Issued by the Basel Committee.” Press Release, January 13, Bank for International Settlements, Basel, Switzerland. http://www.bis.org/press/p110113.htm.

———. 2011b. “Basel III: “A Global Regulatory Framework for More Resilient Banks and Banking Systems.” BCBS Publica-tion No. 189, June, Bank for International Settlements, Basel, Switzerland. http://www.bis.org/publ/bcbs189.htm.

———. 2011c. “Outcome of the September 2011 Basel Commit-tee Meeting.” Press Release, September 28, Bank for Interna-tional Settlements, Basel, Switzerland. https://www.bis.org/press /p110928.htm.

———. 2011d. “Basel III Definition of Capital Frequently Asked Questions.” BCBS Publication No. 204, December, Bank for

International Settlements, Basel, Switzerland. http://www.bis .org/publ/bcbs204.htm.

———. 2012. “Peer Review of Supervisory Authorities’ Imple-mentation of Stress Testing Principles.” BCBS Publication No. 204, December, Bank for International Settlements, Basel, Switzerland. http://www.bis.org/publ/bcbs218.htm.

———. 2013a. “Regulatory Consistency Assessment Programme (RCAP)—Analysis of Risk- weighted Assets for Market Risk.” BCBS Publication No. 240, January, Bank for International Set-tlements, Basel, Switzerland. https://www.bis.org/publ/bcbs240 .htm.

———. 2013b. “Regulatory Consistency Assessment Programme (RCAP)—Analysis of Risk- weighted Assets for Credit Risk in the Banking Book.” BCBS Publication No. 256, July, Bank for International Settlements, Basel, Switzerland. https://www.bis .org/publ/bcbs256.htm.

———. 2013c. “Basel III Phase- in Arrangements.” January, Bank for International Settlements, Basel, Switzerland.

———. 2014. “Capital Floors: The Design of a Framework Based on Standardised Approaches.” BCBS Publication No. 306, Consultative Paper, December, Bank for International Settle-ments, Basel, Switzerland. https://www.bis.org/bcbs/publ/d306 .htm.

———. 2016a, “Minimum Capital Requirements for Market Risk.” BCBS Publication No. 352, January, Bank for Interna-tional Settlements, Basel, Switzerland. https://www.bis.org /bcbs/publ/d352.htm.

———. 2016b, “Standardised Measurement Approach for Opera-tional Risk.” BCBS Publication No. 424, March, Bank for In-ternational Settlements, Basel, Switzerland. https://www.bis .org/bcbs/publ/d355.htm.

———. 2016c. “Regulatory Consistency Assessment Programme (RCAP)—Analysis of Risk- Weighted Assets for Credit Risk in the Banking Book.” BCBS Publication No. 363, April, Bank for International Settlements, Basel, Switzerland. https://www .bis.org/bcbs/publ/d363.htm.

———. 2017a. “Basel III: Finalising Post- Crisis Reforms.” BCBS Publication No. 424, December, Bank for International Settle-ments, Basel, Switzerland. https://www.bis.org/bcbs/publ /d424.htm.

———. 2017b. “The Regulatory Treatment of Sovereign Expo-sures.” BCBS Publication No. 425, December, Bank for Inter-national Settlements, Basel, Switzerland. https://www.bis.org /bcbs/publ/d425.htm.

———. 2019, “Minimum Capital Requirements for Market Risk.” BCBS Publication No. 457, January, Bank for Interna-tional Settlements, Basel, Switzerland. https://www.bis.org /bcbs/publ/d457.htm.

Boston Consulting Group. 2011. Risk Report 2011: Facing New Re-alities in Global Banking. Boston Consulting Group, Bethesda, Maryland.

Bruno, Brunella, Giacomo Nocera, and Andrea Resti, 2015. “The Credibility of European Banks’ Risk- Weighted Capital: Struc-tural Differences or National Segmentations?” BAFFI CARE-FIN Centre Research Paper No. 2015-9, June 3. https://ssrn .com/abstract=2613943.

de Longevialle, Bernard, and Thierry Grunspan. 2011. “Despite Sig-nificant Progress, Capital Is Still a Rating Weakness for Large Global Banks.” Standard & Poor’s, January 18. http://www .alacrastore.com/ s- and- p- credit- research/ Despite- Significant - Progress- Capital- Is- Still- A- Rating- Weakness- For- Large- Global - Banks- 844003.


https://www.bankofengland.co.uk/-/media/boe/files/financial-stability-report/2011/december-2011.



https://www.bankofengland.co.uk/-/media/boe/files/news/2011/december/



https://www.bis.org/publ/bcbs04a.htm.

https://www.bis.org/publ/bcbs04a.htm.






https://www.bis.org/bcbs/irbriskweight.htm.

http://www.bis.org/publ/bcbs118.htm.




http://www.bis.org/publ/bcbsca.htm.

http://www.bis.org/publ/bcbs/basel2enh0901.htm.

http://www.bis.org/press/p110113.htm.


https://www.bis.org/press/p110928.htm.

https://www.bis.org/press/p110928.htm.






















https://ssrn.com/abstract=2613943.de

https://ssrn.com/abstract=2613943.de

http://www.alacrastore.com/



www.bankofengland.co.uk/-/media/boe/files/quarterly-bulletin /2011/ boe- speeches- 2011-q2.

Hannoun, Hervé. 2011. “Sovereign Risk in Bank Regulation and Supervision: Where Do We Stand?” Financial Stability Insti-tute High- Level Meeting, Abu Dhabi/UAE, October 26. https://www.bis.org/speeches/sp111026.htm.

International Financial Reporting Standards (IFRS) Foundation. 2018. “Use of IFRS Standards Around the World.” London, September.

International Monetary Fund (IMF). 2012. Global Financial Sta-bility Report, Chapter 3. Washington, DC, April. https://www .imf.org/~/media/Websites/IMF/ imported- f lagship- issues /external/pubs/ft/GFSR/2012/01/pdf/_c3pdf.ashx.

Jobst, Andreas, and Anke Weber. 2016. “Profitability and Balance Sheet Repair of Italian Banks.” IMF Working Paper No. 16/175, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 / Profitability- and- Balance- Sheet-Repair-of-Italian-Banks -44195.

Le Leslé, Vanessa, and Sofiya Avramova. 2012. “Revisiting Risk- Weighted Assets.” IMF Working Paper No. 12/90, International Monetary Fund, Washington, DC. https://www.imf.org/en /Pub lications/WP/Issues/2016/12/31/Revisiting-Risk-Weighted - Assets- 25807.

Office of the Comptroller of the Currency/Board of Governors of the Federal Reserve System/Federal Deposit Insurance Corporation (OCC/Federal Reserve/FDIC). 2012. “Regula-tory Capital Rules: Regulatory Capital, Implementation of Basel III, Minimum Regulatory Capital Ratios, Capital Ade-quacy, Transition Provisions, and Prompt Corrective Action.” Federal Register 77 (169), August 30.

Masters, Brooke, and Ed Hammond. 2011. “FSA Crackdown on Understated Property Risks.” Financial Times, December 11. https://www.f t.com/content/340ec5ee-23f3-11e1-a4b7 -00144feabdc0.

Morgan, Don, and Adam B. Ashcraft. 2003. “Using Loan Rates to Measure and Regulate Bank Risk: Findings and an Immodest Proposal.” Journal of Financial Services Research 24 (2-3), 181–200.

Pandit, Vikram. 2012. “Apples vs. Apples— A New Way to Mea-sure Risk.” Financial Times, January 10. https://www.ft.com /content/90bb724 a- 3 afc- 11e1-b7 ba- 00144feabdc0.

Samuels, Simon, and Mike Harrison. 2011a. “Two Hundred Mil-lion Inputs— Can You Trust Risk Weightings at European Banks?” Barclays Capital, April 6.

———. 2011b. “Saving Risk Weightings— How Can European Banks Regain Trust in Risk Weighting?” Barclays Capital, July 12.

Santos, André O., and Douglas Elliott. 2012. “Estimating the Costs of Financial Regulation.” IMF Staff Discussion Note 12/1, Inter-national Monetary Fund, Washington, DC. https://www.imf.org /en/Publ icat ions/ Sta f f- Discussion- Notes/Issues/2016 /12/31/ Estimating-the-Costs-of-Financial- Regulation- 26231.

Tarullo, Daniel. 2011. Testimony before the Committee on Finan-cial Services, US House of Representatives. June 16. http://www .federalreserve.gov/newsevents/testimony/tarullo20110616a.htm.

Watt, Michael. 2011. “FSA’s Turner: RWA Divergence Would Undermine Basel III,” Risk.Net, June 24. http://www.risk.net/ risk-magazine/news/2081533/fsas-turner-rwa-divergence - undermine- basel- iii.

de Longevialle Bernard, Elie Hériard- Dubreuil, and Thierry Grunspan. 2008. “Towards Comparable Basel II Ratios: Stan-dard & Poor’s Risk- Adjusted Capital Framework.” In Pillar II in the New Basel Accord— The Challenge of Economic Capital, edited by A. Resti. London: Risk Books, 397–422. http://riskbooks .com/ pillar- ii- in- the- new- basel- accord.

Demirgüç- Kunt, Asli, Enrica Detragiache, and Ouarda Mer-rouche. 2010. “Bank Capital: Lessons from the Financial Cri-sis.” Policy Research Working Paper No. 5473, The World Bank, Washington, DC. http://documents.worldbank.org /cu rated/en/568301468325454646/Bank-capital-lessons-from - the- financial- crisis.

Dodd- Frank Wall Street Reform and Consumer Protection Act of 2010, Pub. L. No. 111-203. https://www.congress.gov/bill/111th -congress/ house- bill/4173/text.

Duffie, Darrell, and Kenneth J. Singleton. 2003. Credit Risk: Pricing, Measurement, and Management. Princeton: Princeton University Press.

European Banking Authority (EBA). 2011. “Capital Buffers for Addressing Market Concerns over Sovereign Exposures– Methodological Note.” October 26, European Banking Authority, London. https://eba.europa.eu/documents/10180/26923/Sovereign - capital- shortfall_ Methodology- FINAL.pdf/acac6c68-398e-4aa2 -b8a1-c3dd7aa720d4.

———. 2018. “ EU- wide Transparency Exercise: Risk Assessment Re-ports.” December 14, European Banking Authority, London. https://eba.europa.eu/ risk- analysis-and-data/eu-wide-transparency -exercise/2018.

European Union. 2013a. “Directive 2013/36/EU of the European Parliament and of the Council on Access to the Activity of Credit Institutions and the Prudential Supervision of Credit Institutions and Investment Firms, Amending Directive 2002/87/EC and Repealing Directives 2006/48/EC and 2006/49/EC (Capital Requirements Directive).” Official Jour-nal of the European Union L 176/338-436 (June 2).

———. 2013b. “Regulation (EU) No. 575/2013 of the European Parliament and of the Council on Prudential Requirements for Credit Institutions and Investment Firms and Amending Regula-tion (EU) No. 648/212 (Capital Requirements Regulation).” Of-ficial Journal of the European Union L 321/6-342 (November 30).

Financial Services Authority (FSA). 2010a. Banks, Sovereigns and Large Corporates HPE: Technical Appendix. Financial Services Authority: London. https://www.yumpu.com/en/document/view /3245890/ banks-sovereigns-and-large-corporates-hpe- technical - appendix.

———. 2010b. Results of 2009 Hypothetical Portfolio Exercise for Sovereigns, Banks and Large Corporations. March 1, Financial Services Authority: London.

Fitch. 2010. Core Capital in Financial Institutions: The Fitch Perspective. March. http://www.fitchratings.com/creditdesk /reports/report_frame.cfm?rpt_id=502106.

———. 2012. Fitch Core Capital: The Primary Measure of Bank Capitalisation Special Report. January. http://www.alacrastore .com/ fitch- credit- research/Fitch-Core-Capital-The-Primary - Measure- of- Bank- Capitalisation- 650111_report_frame.

Gordy, Michael B., and Marrone James. 2010. “Granularity Ad-justment for Mark- to- Market Credit Risk Models.” Finance and Economics Discussion Series 2010-37, Federal Reserve Board, Washington, DC. https://papers.ssrn.com/sol3/papers .cfm?abstract_id=1895495##

Haldane, Andrew. 2011. “Capital Discipline.” Keynote Speech, American Economic Association, Denver, January 9. https://


http://www.bankofengland.co.uk/-/media/boe/files/quarterly-bulletin/2011/

http://www.bankofengland.co.uk/-/media/boe/files/quarterly-bulletin/2011/

https://www.bis.org/speeches/sp111026.htm

https://www.imf.org/~/media/Websites/IMF/




https://www.imf.org/en/Pub


https://www.ft.com/content/340ec5ee-23f3-11e1-a4b7-00144feabdc0

https://www.ft.com/content/340ec5ee-23f3-11e1-a4b7-00144feabdc0

https://www.ft.com/content/90bb724

https://www.ft.com/content/90bb724



http://www.federalreserve.gov/newsevents/testimony/tarullo20110616a.htm

http://www.federalreserve.gov/newsevents/testimony/tarullo20110616a.htm

http://www.risk.net/

http://riskbooks.com/

http://riskbooks.com/

http://documents.worldbank.org/cu


https://www.congress.gov/bill/111th-congress/

https://www.congress.gov/bill/111th-congress/

https://eba.europa.eu/documents/10180/26923/Sovereign-capital-


https://eba.europa.eu/

https://www.yumpu.com/en/document/view/3245890/


http://www.fitchratings.com/creditdesk/reports/report_frame.cfm?rpt_id=502106.

http://www.fitchratings.com/creditdesk/reports/report_frame.cfm?rpt_id=502106.



https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1895495##












CHAPTER 11

A New Heuristic Measure of Fragility and Tail Risks: Application to Stress Testing

NASSIM NICHOLAS TALEB • ELIE CANETTI • TIDIANE KINDA • ELENA LOUKOIANOVA • CHRISTIAN SCHMIEDER

This chapter presents a simple heuristic measure of tail risk, which is applied to individual bank stress tests and to public debt. Stress testing can be seen as a first- order test of the level of potential negative outcomes in response to tail shocks. However, the results of stress testing can be

misleading in the presence of model error and the uncertainty attending parameters and their estimation. The heuristic can be seen as a second- order stress test to detect nonlinearities in the tails that can lead to fragility, that is, provide additional information on the robustness of stress tests. It also shows how the measure can be used to assess the robustness of public debt forecasts, an important issue in many countries. The heuristic measure outlined here can be used in a variety of situations to ascertain an ordinal ranking of fragility to tail risks.

This chapter stakes out an intermediate ground, showing how existing stress testing models can be used to develop a simple measure of tail risk by taking into account convexity effects. Economic and financial models have limitations and constraints, including misspecification, estimation using as-sumed and sometimes inaccurate probability distributions, and so on. As a result, using such models to estimate the potential impact of shocks may lead to increasingly inaccu-rate estimates the further in the tails such shocks are. Even when testing is undertaken to try to detect the sensitivity of an outcome to different sized shocks (sensitivity testing), the focus tends to be on the range of possible levels of outcomes.

This chapter applies to stress testing a simple heuristic method proposed by Taleb 2011, based on methods to de-tect hidden exposures to volatility in option- trading portfo-lios. This method allows for evaluating how well tail risks are captured by stress tests. Rather than using point estimates (or ranges) of outcome levels, this method calculates the dif-ference between outcomes to look for potential convexities

1. INTRODUCTIONMuch of the history of the financial crisis can be interpreted broadly as an underestimation of risks, not only of the prob-ability of “black swans” ( large- impact, unforeseen, random events), but of the financial system’s fragility to them. For example, the potential for bank losses and disruption of bank funding markets due to deterioration in the US sub-prime housing market and the potential for widespread sov-ereign and banking distress in the euro area triggered by stresses in Greek sovereign finances were both severely underappreciated.1

A great deal of soul searching since has centered around excessive reliance on financial and economic models that are seen to have led policymakers and financial markets astray, in part by giving an unwarranted level of confidence about the potential size of downside outcomes. Most of this effort has been aimed at, on the one hand, scrapping models alto-gether, or on the other, seeking to design and develop better models.

This chapter is based on IMF Working Paper 12/216 (Taleb and others 2012). It benefited from comments by Gianni De Nicolo and Christopher Towe.1 Arguably, these were not black swan events, as both subprime losses and Greek sovereign distress were in principle foreseeable, but in both cases, the

magnitude of the outcomes would at least have been regarded as fairly extreme tail events a year or two prior to their full flowering.


A New Heuristic Measure of Fragility and Tail Risks: Application to Stress Testing262

firm or government can be exposed to the underestimation of a certain set of tail risks and, what is critical, how vulnerable it can be to model error.

The chapter is organized as follows: Section 2 first pro-vides some concepts and methods to assess fragility in general and then elaborates the proposed heuristic. Section 3 presents two case studies applying the heuristic to the outcome of bank and fiscal stress tests, respectively. Section 4 gives an overview of how the heuristic could be used in stress testing applications. Section 5 concludes.

2. REVIEW OF CONCEPTS TO ASSESS FRAGILITY

The Current State of Stress Testing

The crisis revealed weaknesses in the stress testing exercises performed on financial systems and institutions in several countries, leading the IMF and country financial regulatory authorities to pay more attention to stress testing and to over-haul existing methodologies. Specifically, the crisis demon-strated that stress tests with poorly designed scenarios omitted shocks, based on inappropriate methods, a narrow coverage of institutions, and so on, can produce results that provide a false reassurance about the degree of financial stability.5

However, there are fundamental issues with stress test-ing, especially seemingly more sophisticated ones: First, many stress tests focus on the point estimates of very few scenarios, and often pay little attention to how the impact would change in case of different scenarios, for example, a slightly more severe one.6 Second, if stress tests do not take into account the possibility of model and parameter error, it can be misleading to rely only on the point estimates of even well- designed stress tests. Without considering the potential for these errors, one could miss the convexities and nonlin-earities that can lead to serious financial fragilities.

The main focus of stress testing has been expanding from solvency and market risk toward liquidity and contagion risks. Compared to solvency stress tests, liquidity stress tests are less developed for several reasons, including (1) that li-quidity risk was seen as “less of a critical issue” until the global crisis; (2) that liquidity crises were seen as low- frequency, but potentially high- impact events; and (3) that to some extent, all liquidity crises were seen as unique.

Contagion- risk stress tests require more elaborated data on individual cross exposures on the interbank market and across broad financial and economic sectors, in order to build up maps of interconnectedness. Collection of such

which, if ignored, might lead stress testers to underestimate (or, likely in fewer cases, overestimate) the impact of tail events. The very simplicity of this heuristic is seen as a vir-tue. It should focus attention on potential non linearities in the tails that are rarely given prominence, and yet should be easily understood and taken into account once made explicit.

These nonlinear (convexity) effects in the tails can cause, for instance, financial losses or sovereign debt unexpectedly to “blow up” in response to shocks that are only a little larger than anticipated and therefore remain invisible. Even if the shocks are correctly foreseen, model error or parameter un-certainty may also lead to substantial underestimation of risks. However, with the heuristic, even though one may not be able to rely on estimated levels of potential losses from a given model, looking at how these levels vary in small ranges around the shock being tested should give a sense of whether the difference in loss estimates is growing (or pos-sibly tamping down) as one moves further out the tail of adverse shocks.

Taleb 2011 gives a simple analogy for how such a second- order test can provide valuable information even when a measuring tool may be flawed. Using an inaccurate tape measure will give a false reading of a child’s height (a level measurement). However, if one uses the same tape measure over time, it will give a reliable test of whether the child is growing (a second- order measurement). By the same token, most economic and financial models have limitations, but looking at the differences in estimated outcomes from any given model will be robust under fairly general conditions, thus pointing the stress tester in the right direction.2

More specifically, the proposed heuristic takes advantage of a version of Jensen’s inequality (applied to higher order terms) to detect convexity in the tails. The idea is to take small, equal- sized perturbations around the results of a tail- risk test and see whether the differences in risk measurements indicate convexity (or linearity or concavity). The degree of convexity can then be used as a direct measure of tail risk.3 In other words, if the estimated models are wrong, the levels of corresponding estimated losses may also be substantially wrong, but the heuristic measure should give a reasonable relative ranking of the “fragility” to such convexities.4

In this sense, this heuristic can be seen as a way to elabo-rate the use of stress testing to develop a more robust measure of relative fragilities, since it should capture the convexity of a loss function in the tails. In sum, the heuristic shows how a

2 These conditions include that loss distributions are monomodal, that the bias in the incorrect model (compared to the true model) does not change signs, and that higher differences do not carry opposite signs.

3 The G20 has called for “the IMF to investigate, develop, and encourage implementation of standard measures that can provide information on tail risks.” See IMF/FSB 2009, recommendation 3.

4 Taleb and Douady (2012) develop a theorem that proves how a nonlin-ear exposure maps into tail sensitivity to volatility and model error and produces a transfer function expressing fragility as a direct result of nonlinearity.

5 The most notable example is in Iceland, where stress tests were per-formed just before that country’ liquidity crisis (Ong and Cihak 2010). See also Borio, Drehmann, and Tsatsaronis 2012 for a critical review of the early warning properties of stress tests.

6 Given the conceivably virtually unlimited number of dimensions that could be covered by stress tests, scenarios will hardly ever be realized precisely as assumed by the stress tests.


Nassim Nicholas Taleb, Elie Canetti, Tidiane Kinda, Elena Loukoianova, and Christian Schmieder 263

the necessity to dump roughly $70 billion of stock index fu-tures over three days upon the discovery of Jerome Kerviel’s rogue trading positions in January 2008.9 The large size of the fire sale relative to the size of the market forced Société Générale to realize a particularly large loss because the sale itself caused a particularly adverse price reaction. Had the stock futures been dumped, instead, in ten $7 billion incre-ments over a period, say, of several weeks or months, the ef-fect on prices may have been relatively small. Instead, the price effects of the rapid forced fire sale caused by the size of Société Générale’s positions meant that the losses resulting from one $70 billion sale were far larger than would have been the losses from 10 sales of $7 billion each.

Negative convexity effects can also be produced by posi-tive (that is, reinforcing) feedback effects resulting from complexity and interconnectedness of markets. A financial institution can likely face small day- to- day price variations with relatively little impact on its overall financial position. But if there is a particularly large price variation in a signifi-cant position, the financial institution may be forced to sell some of the position (for example, in order to meet capital requirements or redemption requests). If that sale causes market prices to fall, then further losses can require the fi-nancial institution (and other financial institutions) to liqui-date even more positions, causing further losses and further liquidations, and so on. As a result, the total impact of a large and sudden price decline may be many orders of mag-nitude larger than the impact of a series of smaller price losses that, over time, amount to the same total price de-cline. Similar examples could be adduced from fiscal dy-namics, where higher debt or rollover requirements lead to higher financing costs that raise debt or rollovers, potentially leading to an out- of- control debt spiral.

data commenced only a few years ago following the global financial crisis.7 Thus, even if some network analysis models are developed, data limitations may prevent assessing conta-gion and interconnectivity risks. However, one of the lessons of this chapter is that, by increasing model complexity, the risks of model error increase, which may render stress testing even more vulnerable to the criticism that losses in response to tail events are likely to be severely misestimated.

In sum, despite various elaborations of stress testing tech-niques in the aftermath of the global financial crisis, as well as rethinking the potential magnitude of shocks, it is still an open question whether tail risks are being captured cor-rectly. Are risks estimated properly in the face of potential model error, parameter stochasticity, and incorrect distribu-tions? How would these estimates respond to a marginal change of the stress scenario?

A Simple Heuristic to Detect Fragility

Following Taleb 2011, the heuristic offered below seeks to detect “biases from missed nonlinearities and detection of these using a single ‘ fast- and- frugal,’ model- free, probability free heuristic.” Imagine a payoff structure as shown in Fig-ure 11.1. With a linear payoff, the “harm” of an adverse shock is proportional to the size of the shock. But a payoff with concavity (negative convexity) becomes disproportion-ately larger as the shock (event size) becomes larger8. With particularly large black- swan- type events, the difference in harm between a linear and negatively convex payoff can es-calate exponentially.

Such negative convexity effects are quite frequent in eco-nomic and financial situations. For instance, they may result from size. The French bank Société Générale was faced with

7 See, for example, FSB 2011.8 Note that if exposure to an event is negative (for example, a short posi-

tion), then the concave payoff structure shown in the diagram would actually become convex, and in fact would be exactly the negative of the concave payoff function.

9 According to an investigation into the scandal by Société Générale’s own General Inspection department, a €49 billion (US$71 billion) long position on index futures was discovered on January 20 and unwound between January 21 and January 23, leading to gross losses of €6.4 bil-lion (see Société Générale 2008).

Black Swan Event

Harmf(X ) Event Size

X

Source: Authors.

Figure 11.1 Why the Concave Is Hurt by Tail Events



How Can the Simple Heuristic Enhance Stress Tests?

The heuristic provides a technique to assess how nonlinear tail risks are, and thereby to assess the sensitivity of the out-come of stress tests vis- à- vis different risk drivers, in a way that is more robust to model error. As macro stress tests are usually based on a very limited number of stress scenarios, the heuristic enhances the scope of macro stress tests with limited additional effort and thereby fills an important gap in the stress testing toolkit.

As shown in Figure 11.1, the outcome of solvency stress tests, measured in terms of changes in capitalization, is usu-ally highly skewed to the left, that is, there is a limited chance that capitalization will go up significantly, while there is considerable risk that capitalization will drop sharply in response to a stress (fat tail). Most macro stress tests ex-plore a very limited “area” of the distribution function, often limited to a baseline scenario and a few stress scenarios, whereby one computes a few point estimates, but the sensi-tivity of the outcome to changes in key risk drivers remains hidden.13

As the distribution tends to be particularly nonlinear in the tails, which is when banks (and systems) come close to the brink of failure, it is essential to understand these non-linearities. Rather than running a series of additional sce-narios, the heuristic allows for testing the sensitivity of the outcome in an efficient and robust manner.14 As such, the heuristic could be used as a standardized method to measure tail risks.

The outcome of the heuristic in the box on the left-hand side of Figure 11.2 is illustrated further in Figure 11.3, which displays the possible nonlinearities that would be tested by the simple heuristic.15

Finally, negative convexity effects may actually be pro-duced by regulation, because of incentives for traders to “hide risks in the tails.” For example, if a regulator requires capital to be set aside against the possibility of an adverse market price move, calculated according to a value at risk model at, say, 5 percent probability, a trade that has a large payoff for a tail shock with a 4 percent probability may fall outside the regulator’s view, thereby providing traders with an asymmetric incentive to place such a trade.10

A simple point estimate from a stress test gives no sense of such potential convexity effects. Such an estimate will be based on a single choice of shock, which can be thought of as applying the model to an average shock in the tails. The heu-ristic involves averaging the model results over a range of shocks. When convexity effects are present, the average of the model results will not be equal to the model results of the av-erage shock. The heuristic is a scalar that measures the extent of that deviation, and is calculated as H, where:

Hf f

fα α α=

− ∆ + + ∆−

( ) ( )

2( ) (11.1)

f(x) is the profit or loss for a certain level α in the state vari-able concerned, or a general vector if one is concerned with higher dimensional cases. ∆ is a change in α, a certain mul-tiple of the mean deviation of the variable. The severity of the convexity expressed by H should be interpreted in rela-tion to the total capital (for a bank stress test, or GDP for a sovereign debt stress debt), and can be scaled by it, allowing for comparability of results, and hence an ordinal ranking of fragilities, among similar types of institutions. When H = 0 (or a small share of the total capital) the outcome is robust, in the sense that the payoff function is linear and the poten-tial gain from a smaller (by the amount ∆) x is equal to the potential loss from an equivalently sized larger x. When H < 0, and significantly so with respect to capital, the out-come is fragile, in the sense that the additional losses with a small unfavorable shock (that is, compared to a given tail outcome) will be much larger than the additional gains with a small favorable shock.11 Thus, volatility is bad in such a situation; it can be said that an institution for which H is negative is “fragile” to higher volatility.12

10 Traders’ compensation systems generally provide extremely asymmetri-cal incentives since traders will receive large bonuses for highly risky trades that pay off, but an equivalent amount cannot be clawed back from them should the trade result in large losses, since their salaries will be bounded at zero. Such asymmetric incentives can lead traders to take on more risk than they would if there were a symmetrical incentive scheme.

11 Note that if the stress test involves something where a larger result rep-resents the adverse case, such as in the change in the net debt/GDP ratio examined in Case Study 2, fragility will be represented by H > 0.

12 Taleb (2012) posits the opposite situation. If one starts with profits, and H > 0, then greater volatility leads to a more profitable outcome than lower volatility. Such a situation is termed “antifragility.” This is not the same as robustness, since with robustness, higher volatility provides nei-ther significant harm nor benefit.

13 For example, the most prominent recent bank stress tests, the Supervi-sory Capital Assessment Program and the subsequent Comprehensive Capital Analysis and Review in the United States, and the two sets of published stress tests conducted by the European Banking Authority used only a baseline scenario and one adverse stress scenario.

14 Of course, ideally, losses would be derived in a closed- form expression that would allow the stress tester to trace out the complete arc of losses as a function of the state variables, but it is exceedingly unlikely that such a closed- form expression could be tractably derived, hence the need for the simplifying heuristic.

15 The mathematical logic that (1) the heuristic reveals tail “fragility,” and (2) that it also reveals model error is demonstrated in Taleb and Douady 2012 as follows. The definition of fragility below a certain level K is the sensitivity of the tail integral— the partial tail expectation— between minus infinity and K to changes in parameters, particularly the lower mean deviation. By the “fragility transfer theorem,” such sensitivity for a variable Y is caused by the second derivative of the function φ, such that Y = φ(x), hence a direct result of the convexity of such function. By the “fragility exacerbation theorem,” the increase in fragility is mapped as a direct effect of such convexity. So it becomes a matter of detecting the convexity, hence the heuristic. Furthermore, parameter impreci-sions in a model are considered as fragilizing if they are capable of caus-ing an increase in the left tail. Taleb and Douady (2012) also show that the convexity bias, that is the misestimation of the effect of Jensen’s in-equality can be obtained by setting K at infinity, to take the effect of the convexity on the total expectation.



the fragility of the result in a single number. Moreover, its very simplicity forces the observer directly to confront the possibility (likelihood?) of error in the level estimates, and the asymmetrical costs of such inaccuracies.

Case Study 1: The Simple Heuristic Applied to Bank Stress Tests

The outcome of a stress test for a sample of 12 large US banks is used to illustrate the functioning of the heuristic. The tests were performed based on the framework developed by Schmieder, Puhr, and Hasan (2011). The stress test was on projected data from the end of 2011, using balance sheet data from the end of 2010, complemented with bank data from the first half of 2011. Stresses were applied for the pe-riod from 2012 to 2016. The outcome of stress was measured in terms of Tier 1 capitalization.

The test projected one single macroscenario, namely a near- zero GDP path with a cumulative deviation from the World Economic Outlook baseline by 10 percentage points during 2012–16. In historical terms for advanced countries

3. THE HEURISTIC APPLIED TO THE OUTCOME OF STRESS TESTS

Purpose for the Use of the Heuristic

As outlined earlier in this chapter, macroeconomic stress tests are usually limited to the computation of a small number of scenarios (for example, the point estimates in Figure 11.2 rep-resented by the vertical lines), which are then used to draw policy conclusions. The drawback of this procedure is that the sensitivities of the outcome to small changes in inputs, although possibly something the team actually running the test may have a feel for, remains hidden for the wider audi-ence, notably including the decision- makers who use the re-sults. Ultimately, the outcome of the scenario is thereby taken at face value. In fact, stress test results are often presented as having a binary “pass/fail” outcome. However, as financial risks are usually highly nonlinear, drawing policy conclusions based on a very few point estimates can produce misleading conclusions. Rather than presenting a series of additional outcomes to policy makers, the simple heuristic summarizes

Change in capitalization(in ppts)

Likelihood

Baseline:+0.5 ppts

Severe stress:–5.0 ppts

Moderate stress:–2.0 ppts

Use of heuristic toexplore shape of

distribution

Distribution function of bank capitalization post scenario

Source: Authors.Note: ppts = percentage points.

Figure 11.2 Illustration of the Use of the Heuristic

HarmEvent Size

f(x )

H

f (x + Δ) + f (x – Δ)2

Source: Authors.

Figure 11.3 Fragile and Antifragile Outcomes of Stress Tests



However, there are also some antifragile outcomes, re-flecting risk- mitigating effects such as a drop of risk- weighted assets (RWAs) (through credit losses), deleveraging that helps to mitigate stress, as well as other factors such as banks’ relative levels of losses and income (that is, risk and return) and other effects. These cases are instances of deleveraging (due to a reduction in assets under stress as a result of credit losses), which has a positive and non linear marginal impact on the risk profile of a bank under stress.20

In addition to providing information on whether the tail is nonlinear as such, the ratio can also provide information on how nonlinear it is, and thereby allows for ordinal com-parisons. The outcome allows for computing the additional impact of a further drop of GDP growth by 1 (or 2) mean absolute deviations on various key drivers of bank capitaliza-tion, for example.

Table 11.2 sheds some more light on the fragility of banks in relative terms. Banks are ordered according to (1) the im-pact of the adverse scenario in terms of capitalization (the outcome of the actual standard stress tests, shown in row 0 in Table 11.1), and the heuristic for; (2) scenarios 2/4; and (3) scenarios 18/20. The latter two cases are assumed to be a proxy for the overall fragility of the banks.

Overall, the rank order tends to be consistent across the three measures, but with several exceptions. Bank 9, for example, is the least vulnerable according to both measures. However, while a “traditional” stress test would single out Bank 6 as vulnerable, the heuristic classifies this bank as a less fragile one. The opposite is true for banks 3, 4, 7, and 12. Intuitively, what this means is that for these banks, their capital would be asymmetrically impacted by negative shocks (compared to similar sized positive shocks), and thus the banks are fragile to a more volatile environment.

Case Study 2: The Simple Heuristic Applied to Public Debt

The heuristic can also be used to predict the effect of debt and the underestimation of the risks of higher- than- planned deficits.

The recent financial crises have highlighted the uncer-tainty around the future path of growth, particularly in ad-vanced economies. Debt and deficit outturns have sometimes been considerably worse than expected, underlining the im-portance of assessing how a worse- than- expected growth scenario over the medium term might impact public finances. This section assesses the sensitivity of public debt to adverse growth shocks for a number of advanced economies.21 More specifically, it analyzes how errors in the estimate of growth shocks could lead to significantly higher error in the estimate of countries’ debt dynamics. Indeed, if nonlin-earities exist, underestimating growth shocks could lead to

(based on the period from 1980 to 2010) this scenario could be expected to occur with a likelihood of about 4 percent.16 The scenario was simulated including a feedback loop be-tween stress in the banking system and macroeconomic growth, that is, banks’ capital needs and the pertinent dele-veraging was simulated to have a negative impact on output, based on Vitek and Bayoumi 2011.17 Together with addi-tional stress elements, especially with respect to losses of trading income and increases of funding costs under stress, the scenario constitutes a highly adverse tail risk scenario.

The impact of macroeconomic stress on key financial risk drivers, namely credit losses, credit growth, and preimpair-ment income, have been estimated using so- called satellite models using panel regressions. The test assumes no recapi-talization other than through retained earnings and allows for deleveraging in case of stress. Further details on the test are provided in Appendix 11.1.

The heuristic was applied to bank capitalization, first, separately for each of the specific risk drivers (credit growth, credit losses, trading income) and then all three together with the impact of GDP growth (Table 11.1).

The outcome indicates that the tail stress test produces nonlinear results in the majority of cases (Table 11.1). For most banks, the outcome is fragile with respect to all of the risk drivers, namely macroeconomic stress (GDP growth change,18 scenarios 1–4), credit growth (5–8), credit losses, (9–12), and income (13–16), as well as to trading income as part of total income (21–24).19 The analysis also includes a combined scenario that stresses credit growth, credit losses, and income by one additional mean deviation (17–20). Fra-gility is especially high for the banks with the worst out-comes, that is, those experiencing the most substantial drop of Tier 1 capitalization under stress (row 0), where all results are fragile (see also Table 11.2). On the other hand, although Banks 3 and 4 would not be found to be the most vulnerable in terms of the impact of stress on capitalization (capitaliza-tion in 2016 versus prestress capitalization, row 0) and the conclusion might therefore be that they are resilient, the heuristic reveals fragilities for those banks.

16 More specifically, scenarios with a cumulative deviation from average growth rates by 10 percentage points (independent from when the de-viation occurs) have been used to compute the likelihood for the occur-rence of the zero- growth scenario.

17 One of the virtues of the heuristic is that it may reveal nonlinearities in the stress testing model, for example, arising from feedback loops, even when such nonlinearities were not explicitly or intentionally built into the stress testing model. This could reveal either a true nonlinearity, or a need to refine the model.

18 GDP affects credit losses, income, and credit growth through the satel-lite models, which makes scenarios 1–4 complementary to scenarios 17–20.

19 By way of a numerical example, the heuristic for Bank 1 under the one standard deviation GDP shock (−0.035) is computed as follows: Under the stress test, the Tier 1 ratio is 1.639 percent, and under additional GDP shocks of +/− one standard deviation is 2.835 and 0.373, respec-tively. Thus, the heuristic is calculated as: (0.373 + 2.835)/2 − 1.639 = −0.035, the same as the computation using the changes of capitalization as shown in Table 11.1: (−1.27 − 1.2)/2.

20 This impact explains the positive H for Banks 5, 8, 9, and 11 in the case of the GDP shock.

21 See Appendix 11.2 for details on the methodology.


Nassim

Nicholas Taleb, Elie C

anetti, Tidiane Kinda, Elena Loukoianova, and Christian Schm

ieder267

TABLE 11.1

The Heuristic Applied to the Outcome of Macroeconomic Stress Tests for the Largest US Banks (Change in T1 capitalization conditional on the scenarios)

Bank 1 Bank 2 Bank 3 Bank 4 Bank 5 Bank 6 Bank 7 Bank 8 Bank 9 Bank 10 Bank 11 Bank 12

Scenario

Impact of adverse scenario −9.60 −13.65 −8.48 −2.82 −8.90 −15.07 −1.61 −0.56 2.27 −18.62 −13.31 −7.63GDP growth 1 MDev worse (1.5 percentage point, cumulative)

−1.27 −2.71 −1.86 −0.80 −0.11 −0.24 −0.71 −0.29 −0.12 −2.01 −1.63 −1.67

GDP growth 2 MDev worse (3.0 percentage points, cumulative)

−2.62 −5.79 −4.01 −1.94 −0.23 −0.48 −1.48 −0.63 −0.24 −4.25 −3.22 −3.49

GDP growth 1 MDev better (1.5 percentage point, cumulative)

1.20 2.52 1.28 0.76 0.11 0.24 0.58 0.31 0.12 1.85 1.68 1.56

GDP growth 2 MDev better (3.0 percentage points, cumulative)

2.34 4.88 2.28 1.48 0.23 0.43 1.16 0.63 0.24 3.67 3.42 3.06

H: +/− 1 MDev −0.035 −0.096 −0.294 −0.019 0.000 −0.001 −0.066 0.007 0.000 −0.080 0.025 −0.053

H: +/− 2 MDev −0.141 −0.451 −0.863 −0.230 0.000 −0.025 −0.162 0.000 0.001 −0.293 0.097 −0.216

Credit growth 1 MDev more (3.4 percentage points)

−0.99 −1.71 −1.40 −0.62 −0.03 −0.09 −0.46 −0.28 −0.02 −1.61 −1.51 −1.34

Credit growth 2 MDev more (6.8 percentage points)

−2.07 −3.63 −2.96 −1.57 −0.06 −0.19 −1.16 −0.57 −0.05 −3.39 −3.13 −2.82

Credit growth 1 MDev lower (3.4 percentage points)

0.90 1.53 0.86 0.47 0.03 0.09 0.38 0.26 0.02 1.46 1.41 1.22

Credit growth 2 MDev lower (6.8 percentage points)

1.73 2.96 1.43 0.91 0.05 0.17 0.73 0.50 0.04 2.71 2.73 2.33

H: +/− 1 MDev −0.043 −0.090 −0.269 −0.076 −0.001 −0.003 −0.039 −0.009 −0.001 −0.076 −0.049 −0.061

H: +/− 2 MDev −0.171 −0.334 −0.764 −0.332 −0.003 −0.011 −0.214 −0.035 −0.003 −0.341 −0.198 −0.244

Credit losses 1 MDev higher (0.11 percentage point)

−1.21 −2.45 −1.68 −0.68 −0.04 −0.13 −0.60 −0.24 −0.02 −2.07 −1.62 −1.55

Credit losses 2 MDev higher (0.22 percentage point)

−2.50 −5.16 −3.59 −1.65 −0.09 −0.26 −1.21 −0.47 −0.04 −4.40 −3.21 −3.25

Credit losses 1 MDev lower (0.11 percentage point)

1.13 2.26 1.16 0.64 0.04 0.13 0.48 0.24 0.02 1.88 1.63 1.45

Credit losses 2 MDev lower (0.22 percentage point)

2.21 4.44 2.04 1.28 0.09 0.26 0.96 0.49 0.04 3.70 3.28 2.83

H: +/− 1 MDev −0.036 −0.097 −0.262 −0.019 0.000 0.000 −0.057 0.002 0.000 −0.094 0.009 −0.052

H: +/− 2 MDev −0.148 −0.360 −0.773 −0.184 0.001 0.000 −0.127 0.010 0.001 −0.353 0.037 −0.210

(continued)


A N

ew H

euristic Measure of Fragility and Tail Risks: A

pplication to Stress Testing268

The Heuristic Applied to the Outcome of Macroeconomic Stress Tests for the Largest US Banks (Change in T1 capitalization conditional on the scenarios)

Bank 1 Bank 2 Bank 3 Bank 4 Bank 5 Bank 6 Bank 7 Bank 8 Bank 9 Bank 10 Bank 11 Bank 12Preimpairment income/TA 1 MDev less (0.17 percentage point)

−1.72 −4.51 −3.44 −2.06 −0.72 −1.17 −1.76 −1.00 −0.96 −1.33 −1.97 −2.66

Preimpairment income/TA 2 MDev less (0.34 percentage point)

−3.68 −10.72 −7.80 −4.95 −1.47 −2.46 −4.55 −2.25 −1.93 −2.76 −3.97 −5.63

Preimpairment income/TA 1 MDev more (0.17 percentage point)

1.55 3.93 2.00 1.56 0.69 0.88 1.31 0.89 0.96 1.24 1.97 2.33

Preimpairment income/TA 2 MDev more (0.34 percentage point)

2.99 6.39 3.67 2.75 1.36 1.53 2.35 1.79 1.93 2.35 3.94 4.29

H: +/− 1 MDev −0.085 −0.291 −0.718 −0.250 −0.014 −0.144 −0.224 −0.051 0.000 −0.043 0.000 −0.162

H: +/− 2 MDev −0.346 −2.163 −2.065 −1.101 −0.058 −0.466 −1.096 −0.230 0.000 −0.202 −0.016 −0.670

All three drivers 1 MDev in adverse direction

−4.34 −10.24 −7.58 −4.45 −0.80 −1.35 −3.86 −1.60 −1.01 −5.63 −5.29 −6.07

All three drivers 2 MDev in adverse direction

−9.47 −33.65 −19.50 −10.92 −1.64 −3.09 −9.95 −4.73 −2.02 −13.11 −11.31 −14.13

All three drivers 1 MDev in benign direction

3.33 6.17 3.38 2.43 0.76 1.03 2.02 1.36 1.00 4.33 4.87 4.23

All three drivers 2 MDev in benign direction

5.17 9.00 6.23 4.19 1.48 1.72 3.67 2.65 2.00 7.91 7.97 6.61

H: +/− 1 MDev −0.505 −2.035 −2.098 −1.012 −0.020 −0.162 −0.924 −0.123 −0.003 −0.649 −0.212 −0.922

H: +/− 2 MDev −2.147 −12.326 −6.638 −3.362 −0.082 −0.685 −3.139 −1.040 −0.012 −2.602 −1.670 −3.756

Trading income/TA 1 MDev less (0.1 percentage point)

−0.48 −1.10 −0.57 −0.26 −0.50 −1.43 −0.25 −0.11 −0.19 −0.49 −0.47 −0.36

Trading income/TA 2 MDev less (0.2 percentage point)

−0.99 −2.12 −1.19 −0.52 −1.03 −3.40 −0.51 −0.20 −0.38 −1.01 −0.95 −0.74

Trading income/TA 1 MDev higher (0.1 percentage point)

0.45 0.97 0.50 0.26 0.48 1.13 0.19 0.11 0.19 0.47 0.47 0.35

Trading income/TA 2 MDev higher (0.2 percentage point)

0.89 1.84 0.98 0.51 0.94 1.96 0.37 0.21 0.38 0.92 0.95 0.69

H: +/− 1 MDev −0.013 −0.063 −0.035 −0.001 −0.012 −0.147 −0.033 0.000 0.000 −0.011 0.000 −0.007

H: +/− 2 MDev −0.053 −0.140 −0.107 −0.005 −0.047 −0.720 −0.070 0.003 0.000 −0.045 0.000 −0.027

Source: Authors.Note: H = heuristic; MDev = mean absolute deviation; TA = total assets; T1 = Tier 1.




pared to smaller or positive growth shocks, indicating a non-linearity (Figure 11.4).

Due to the nonlinearity, there is a disproportionately higher cost to underestimating the growth shock than overes-timating it.22 As illustrated in Table 11.3, the nonlinearities can be important. Thus, when the stress tester is uncertain about the appropriate size of tail shock to choose, for exam-ple, because growth is particularly volatile, symmetrical stress tests around the chosen central shock could help shed light on the impact of higher volatility on debt dynamics.

4. HOW TO APPLY THE SIMPLE HEURISTIC IN IMF STRESS TESTSThe heuristic can be used as a standard element in IMF bank and public debt stress test analysis, for example, as displayed in Figure 11.5. The heuristic can explore potential nonlinearities in some range around the typically sized stress test

disproportionately higher underestimation of countries’ cor-responding debt levels.

Various growth shock scenarios are considered in order to analyze the nonlinearity. A central scenario assumes that growth is 2 percentage points less than in the IMF 2011 per year between 2012 and 2016. This implies a near zero real growth scenario for most of the countries in the sample. In addition to this central scenario, four tail risk scenarios are considered to assess the nonlinearity and thus the fragility of the results. These tail risk scenarios assume that growth is one or two mean deviations above or below the central sce-nario. The mean deviations are estimated over the period 1981–2010 for each country.

The results indicate that the impact of tail growth shocks on debt is nonlinear in all cases, implying that all outcomes are fragile (Table 11.3). Based on a diverse sample of coun-tries, Table 11.3 illustrates changes (in percentage points of GDP) in net debt as a result of tail growth shocks. The sam-ple includes countries with both low and high initial debt levels, countries with low trend growth, countries under market pressure, and countries with large automatic stabiliz-ers. The results illustrate that large negative growth shocks have a disproportionately higher impact on net debt com-

TABLE 11.2

Overall Fragility of BanksBank 1 Bank 2 Bank 3 Bank 4 Bank 5 Bank 6 Bank 7 Bank 8 Bank 9 Bank 10 Bank 11 Bank 12

Rank order (impact of shock based on stress test)

5 3 7 9 6 2 10 11 12 1 4 8

Rank order based on heuristic (scenarios 2, 4)

7 2 1 4 10 8 6 9 11 3 12 5

Rank order based on heuristic (scenarios 18, 20)

7 1 2 4 11 10 5 9 12 6 8 3

Source: Authors.Note: The lower the rank, the higher the impact of stress.

22 Under the hypothesis that the impact of a growth shock on debt is lin-ear, the impact should be similar when a growth shock is augmented and reduced by a similar constant.

TABLE 11.3

Change in Net Debt under Various Scenarios (Percentage points of GDP)

Country A Country B Country C Country D Country E Country F Country G

Central shock (2 percent-age points less)

26.2 13.5 20.9 29.6 28.1 34.5 23.7

Fragility/Antifragility Tests: Change Relative to the Central Shock

GDP growth 1 MDev less 14.0 13.1 16.1 18.5 28.9 39.6 37.5GDP growth 2 MDev less 28.6 27.1 33.2 37.9 60.2 83.6 80.1GDP growth 1 MDev more −13.5 −12.2 −15.2 −17.6 −26.7 −36.0 −33.3GDP growth 2 MDev more −26.4 −23.7 −29.5 −34.3 −51.5 −68.7 −63.1

H: +/− 1 MDev 0.3 0.4 0.5 0.4 1.1 1.8 2.1

H: +/− 2 MDev 1.1 1.7 1.9 1.8 4.4 7.4 8.5

Source: Authors.Note: Since the change in the net debt/GDP ratio is presented in positive numbers, a positive heuristic implies an increase in risk. MDev = mean absolute deviation.



∆X is an estimated mean deviation of the variable X(t) over the pre determined time period.

• Third, compute the outcome for these two additional scenarios.

• Fourth, compute the heuristic accordingly.• Fifth, draw conclusions and reiterate:

- If the heuristic indicates fragility to higher vola-tility (positive when a higher outcome is ad-verse, negative when a lower outcome is adverse), then the stress tester would conclude that an even greater adverse stress could make the out-come substantially worse. The test can also al-low for a rank ordering of fragility to specific risk drivers. This would serve as an additional measure of riskiness beyond the typical “level”

(often a stress of two standard deviations in the state variable). The size of the delta in the stress test, which deter-mines the size of that range, can be chosen with a view to exploring where the stress tester may suspect that nonlineari-ties could arise.23 The basic procedure would be as follows:

• First, run stress tests and obtain results (bank capital ratios, liquidity ratios, nonperforming loan ratios, net interest income, return on assets, return on eq-uity, public debt, and so on).

• Second, take the stress test scenario and construct two additional scenarios: Xt + ∆X, Xt − ∆X, where

BaselineCentral shock+/–2 Md shock

BaselineCentral shock+/–2 Md shock

70

1501. Country C

2011 1612 13 1514

Gene

ral G

over

nmen

t Gro

ss D

ebt (

In p

erce

nt o

f GDP

)

100

3002. Country F

2011 1612 13 1514

Gene

ral G

over

nmen

t Gro

ss D

ebt (

In p

erce

nt o

f GDP

)

90

110

130

150

200

250

Source: Authors.Note: Md = mean absolute deviation.

Figure 11.4 Illustration of Debt Dynamics under Various Scenarios

Step 1: Run stress test and obtain result for a few scenarios

Step 2: Construct additional scenariosby changing one or more risk parameters/inputs by x standard deviations in both directions

Step 3: Compute outcome for the additional scenarios

Step 4: Compute heuristic

Step 5: Draw conclusions and reiterate (optional)

Source: Authors.

Figure 11.5 The Simple Heuristic as an Integral Part of Stress Test Frameworks

23 For example, if the nonlinearity arises from traders hiding risks in the tails (as posited in Section II.B), then a relatively small delta plus and minus around the 5 percent probability level could reveal such a nonlinearity.



represented by the state variables), with fragility defined as the property that stresses bring disproportionately higher harm as the stress increases.

The heuristic is calculated by conducting additional stress tests around the central stress scenario, by varying the rele-vant state variable (or vector), plus and minus, by a multiple of its mean deviation. The heuristic itself is a simple scalar that can be easily understood.

This paper constructs heuristics for two different stress testing applications, bank capitalization, and public sover-eign debt. In both cases, there are a priori reasons to be con-cerned about dynamics that could give rise to nonlinearities of the type that can cause fragility. Indeed, in both sets of stress tests, we find cases in which such nonlinearities appear. This finding should lead the stress tester toward a more cau-tious interpretation of the robustness of the results of a single point stress test, or consider running more scenarios. The heuristic, as a kind of second- order stress test, may well give a different ordinal ranking than would be found by examin-ing only the “level” of stress found in the point test. That is, the heuristic presented in this paper could lead to different (or at least additional) conclusions to the pass/fail results that are typically presented as the main outcome of stress tests. Thus, the heuristic explicitly highlights the potential for harm (or conceivably benefit) from high volatility of stresses in the tails.

Such results may have important policy implications. For example, if a country finds the structure of its public fi-nances makes it particularly fragile to growth shocks, it may conclude it has less room for countercyclical deficits. By the same token, a stress test could help a bank or a banking super-visor find otherwise hidden fragilities coming from, say, large illiquid positions subject to fire sales in some conditions, de-rivative exposures with nonlinear payoffs, or feedback loops between losses and funding costs.

More broadly, the heuristic could be of quite general applica-tion whenever a system is being subjected to a stress test. Cal-culation and presentation of the heuristic, in particular, could become a useful standard statistic in stress testing applica-tions, as well as a diagnostic tool that could be used to sug-gest when the stress tester may need to dig deeper to explore the robustness of the results of the original stress test.

stress test, and furthermore, one that is more ro-bust to some types of model errors.

- A finding of fragility to higher volatility could suggest reiterating the procedure in order to ex-plore further how outcomes can vary in differ-ent parts of the tail. That is, additional scenarios with smaller or larger adverse shocks could be used to explore other areas of the risk distribu-tion function. In that case, one could re iterate from Step 1 (a comprehensive reiteration with a different set of stresses) or Step 2 (a limited reit-eration choosing different ∆s).24

5. CONCLUSIONThis chapter has presented a conceptually and operationally simple heuristic, developed in Taleb 2011, to expand infor-mation that can be extracted from the results of stress tests. When there is uncertainty about the stress testing model or the potential size of tail shock to be tested, as will generally be the case, the level results, generally presented in a simple pass/fail form, will be imprecise in some measure (as stress testers themselves will be aware). By examining the convex-ity of losses in the tails of the distribution, the stress tester can gain an extra order of information on the fragility of what is being tested. These convexities can be important, since they can cause financial losses, sovereign debt, or some other finan-cial variable to “blow up” in response to a shock that may be only modestly larger than analyzed by the stress tester.

The heuristic will also be important in the way stress test results are presented. It provides an intuitively understood measure of the bias stemming from the likely imprecision necessarily involved in stress testing. Alternatively, it can be seen as a measure of fragility to volatility in the stresses (as

24 There will be no obvious procedure to specify when the iterations should stop since the functional form of the relationship between out-comes and stressors may be almost limitlessly complex. Rather, the it-eration of the procedure will be used to expand the stress tester’s knowledge about behavior in the tails, but the stress tester should never have a pretense to complete knowledge of the functional form govern-ing how the outcome will react to different stressors.


Appendix 11.1.Details on Macroeconomic Bank

Stress Test

BANK- SPECIFIC CREDIT LOSSESThe Japan- like growth shock (that is, a lost- decade- like scenario) results in substantial losses in banks’ loan books as well as on assets subject to counterparty credit risk, which were assumed to change in line with loan loss rates (in relative terms, but on a substantially lower level).25 A house price shock that results in an additional “permanent” decrease of house prices by 20 percent is added, leading to loss given default (that is, 1 minus the recovery rate) of 40 percent for retail mortgages.26

The stress test also simulated worst- case trading results during 2012 and 2013, in line with very severe historical cases. Spe-cifically, trading income is projected to yield losses of around 1.5 percent of total assets in 2012 (in line with adverse levels ob-served in recent years leading up to that point) and of around 1 percent in 2013.27

CREDIT GROWTH AND RISK- WEIGHTED ASSETS Credit growth is estimated through satellite models, reflecting GDP growth. Credit growth becomes slightly negative, which gives banks some breathing space to digest losses.

Risk- weighted assets (RWAs) grow in line with total credit exposure (accounting for credit growth on the one hand and losses on the other), while risk effects (as would be the case under the internal- ratings- based approach) are disregarded, as US banks were not required to use the internal- ratings- based approach during this period. Likewise, it is assumed that there is no behavioral adjustment to benefit from lower RWAs (by replacing loans by securities, for example), which is a conservative as-sumption, as banks can be expected to change their asset profiles during a five- year period if they need to free up RWAs. The remainder of the RWAs (that is, for market risk and credit risk) are held constant.

INCOME AND RETAINED EARNINGSThe banks’ preimpairment pretax income28 under the adverse growth scenario was projected, using a satellite model, to drop 13 percent compared to 2010, and to remain at that level throughout the projection period, reflecting the quasi–zero growth path.29 It was assumed that 60 percent of net income is retained (in line with empirical evidence) if net income remains posi-tive, otherwise retained income (that is, loss) is fully retained. A tax rate of 25 percent was applied to all banks, in line with empirical evidence.

FUNDING COSTSA funding cost increase was assumed conditional on the capitalization level (during stress), that is, funding costs increase faster as capital falls to low levels, reflecting empirical evidence. Hence, the funding costs add a dynamic element to the stress test. It was assumed that banks are able to pass on (only) 50 percent of funding cost increases to their customers, given competition pressures. Further information is given in Schmieder and others 2012.

25 The mean deviation has been computed based on the evolution of the median loss rate among all US banks in Bankscope.26 20 percent is a common, conservative benchmark for housing loans (many banks use figures of about 15 percent).27 The loss for a specific bank will depend on the intensity of its trading business.28 The preimpairment income includes all sources of operating income other than trading income, that is, interest income, commission and fee income, and

so on.29 The 2010 preimpairment pretax income was very close to the average over the last 5 years except for investment banks, where trading income was more

important.


Appendix 11.2.Details on Public Debt Stress Test

The following standard debt dynamic equations illustrate the main parameters influencing debt dynamics:

(1 )

1

( )

1d d r pb

ri g

g

pb pb og

t t t t

t

t tbase

R E tε ε

= + −

= −+

= + − ∆

−

dt represents the ratio of public debt to GDP, pbt is the primary balance, and rt is the growth-adjusted interest rate. The growth- adjusted interest rate is a function of the nominal interest rate (i) and the nominal GDP growth rate (g). The nominal interest rate is derived from the ratio of interest payments during the current year to the end- period stock of debt during the previous year. The primary balance (pbt) depends on the primary balance under the baseline (pbt

base), revenue and expenditure semi- elasticity to changes in the output gap (εR, εE),30 and the change in the output gap between the baseline and different scenarios (∆ogt). All scenarios assume that growth shocks do not affect potential GDP and governments do not take any discretionary corrective measures to smooth their impacts.31 As a consequence, growth shocks affect debt ratios through the size of automatic stabilizers and changes in the GDP base.

Three macroeconomic variables will therefore affect each country’s debt dynamics: trend growth, the size of the initial (pre-shock) stock of public debt, and the size of the automatic stabilizers. Trend growth would particularly matter for countries with projected low growth rates during the period 2012–16. For these countries, a negative growth shock would lead to a signifi-cantly higher buildup of public debt than in high- growth countries. The initial stock of public debt would be particularly im-portant for highly indebted countries, notably those that experienced a surge in their debt ratios as a result of the crisis. The size of automatic stabilizers matters more in countries with particularly high welfare spending, as the relationship between tax revenues and economic activity tends not to vary greatly across countries.

30 Revenue and expenditure elasticity to the output gap are from Girouard and André 2005. When not available, an elasticity of one is assumed for revenue and zero for expenditures.

31 This is a partial equilibrium simulation that also assumes no change in the nominal interest rate as a result of the growth shock.



REFERENCESBorio, Claudio, Mathias Drehmann, and Kostas Tsatsaronis.

2012. “Stress Testing Macro Stress Testing: Does It Live up to Expectations?” BIS Working Paper 369, Bank for International Settlements, Basel, Switzerland. https://www.bis.org /publ/work369.htm.

Financial Stability Board (FSB). 2011. “Understanding Financial Linkages: A Common Data Template for Global Systemically Important Banks.” Consultation Paper, Financial Stability Board, London. http://www.fsb.org/2011/10/r_111006/.

Financial Stability Board (FSB) and International Monetary Fund (IMF). 2009. The Financial Crisis and Information Gaps. Report to the G20 Finance Ministers and Central Bank Governors, Financial Stability Board, London. http://www.fsb.org/2009/10 /r_091029/.

Girouard, Nathalie, and Christophe André. 2005. “Measuring Cy-clically Adjusted Budget Balances for OECD Countries.” OECD Economics Department Working Paper 434, Organi-sation for Economic Co-operation and Development, Paris. https://www. oecd- ilibrary.org/economics/ measuring- cyclically - adjusted-budget-balances-for-oecd-countries_787626008442.

International Monetary Fund (IMF). 2011. World Economic Outlook— Slowing Growth, Rising Risks. Washington, DC, September. https://www.imf.org/en/Publications/WEO/Issues / 2016/12/31/Slowing-Growth-Rising-Risks.

Ong, Li Lian, and M. Cihak. 2010. “Of Runes and Sagas: Perspec-tive on Liquidity Stress Testing Using an Iceland Example.” IMF Working Paper 10/156, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ Of- Runes- and- Sagas- Perspectives- on -Liquidity- Stress- Testing-Using-an-Iceland-Example-24019.

Schmieder, Christian, Heiko Hesse, Benjamin Neudorfer, Claus Puhr, and Stefan W. Schmitz. 2012. “Next Generation System- Wide Liquidity Stress Testing.” IMF Working Paper 03/12, Interna-tional Monetary Fund, Washington, DC. https://www.imf .org/external/pubs/cat/longres.aspx?sk=25509.0.

Schmieder, Christian, Claus Puhr, and Maher Hasan. 2011. “Next Generation Balance Sheet Stress Testing.” IMF Working Paper 11/83, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 / Next- Generation-Balance-Sheet-Stress-Testing-24798.

Société Générale. 2008. “Mission Green: Summary Report.” So-ciété Générale, General Inspection Department. Paris.

Taleb, Nassim N. 2011. “A Map and Simple Heuristic to Detect Fragility, Antifragility, and Model Error.” NYU- Poly Working Paper, New York University, New York.

———. 2012. Antifragile: Things That Gain from Disorder. New York: Random House; London: Penguin.

———. Elie Canetti, Tidiane Kinda, Elena Loukoianova, and Christian Schmieder. 2012. “A New Heuristic Measure of Fragility and Tail Risks: Application to Stress Testing.” IMF Working Paper 12/216, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP /Issues /2016 /12/31/ A- New- Heuristic- Measure- of- Fragility- and- Tail - Risks-Application-to-Stress-Testing-26222.

Taleb, Nassim N., and Raphael Douady. 2012. “Mathematical Definition and Mapping of (Anti)Fragility.” NYU- Poly Work-ing Paper, New York University, New York.

Vitek, Francis, and Tamim Bayoumi. 2011. “Spillovers from the Euro Area Sovereign Debt Crisis: A Macroeconometric Model- Based Analysis.” CEPR Discussion Paper 8497, Center for Eco-nomic Policy and Research, Washington, DC. https://cepr.org /active/publications/discussion_papers/dp.php?dpno=8497.




http://www.fsb.org/2011/10/r_111006/

http://www.fsb.org/2009/10/r_091029/

http://www.fsb.org/2009/10/r_091029/

https://www.imf.org/en/Publications/WEO/Issues/

https://www.imf.org/en/Publications/WEO/Issues/



https://www.imf

http://.org/external/pubs/cat/longres.aspx?sk=


https://www.imf.org/en/Publications/WP

https://cepr.org/active/publications/discussion_papers/dp.php?dpno=8497


https://www. oecd- ilibrary.org/economics/ measuring- cyclically- adjusted-budget-balances-for-oecd-countries_787626008442

https://www. oecd- ilibrary.org/economics/ measuring- cyclically- adjusted-budget-balances-for-oecd-countries_787626008442





CHAPTER 12

How to Capture Macro- Financial Spillover Effects in Stress Tests

HEIKO HESSE • FERHAN SALMAN • CHRISTIAN SCHMIEDER

One of the challenges of financial stability analysis and bank stress testing is how to establish scenarios with meaningful macro- financial link-ages, that is, taking into account spillover effects and other forms of contagion. This chapter presents an approach to simulate the potential

impact of spillover effects based on the “traditional” design of macroeconomic stress tests. Specifically, the chapter examines spillover effects ob-served during the financial crisis and simulates their impact on banks’ liquidity and capital positions. The outcome suggests that spillover effects have a highly nonlinear impact on bank soundness, both in terms of liquidity and solvency.

model; second, vector autoregressive methods; and third, pure statistical approaches. The satellite models commonly take the form of (panel) regression models. The “direct ap-proach” is based on projections of the actual solvency and liquidity parameters without an explicit link to the state of economic and financial variables. While this approach could be equally meaningful in terms of the outcome of stress tests, it does not allow for a detailed storytelling and can underestimate the importance of nonlinear macro- financial factors for bank- specific stress tests.

Modeling contagion effects and their impact typically constitutes a challenge (see Jobst, Ong, and Schmieder 2013, for example). By definition, spillover effects and other dy-namic contagion effects are implicitly captured in past data but not necessarily if one uses structural econometric models— usually perceived as being a “best practice.” Even if potential spillover events are captured in past data, this data might not be representative for a future scenario if, for example, linkages between economies and banks have be-come gradually more intense over time. This study focuses on spillover effects originating from the recent sovereign debt crisis. Other spillover catalysts could be, for instance, a macroeconomic downturn in a major world economy as well as the failure of a large financial institution such as in the case of Lehman Brothers.

1. INTRODUCTIONStress testing has garnered broad attention during recent years, which has spurred numerous conceptual develop-ments.1 Yet, overarching approaches to establish macro- financial linkages, and explicitly capture the nonlinearity of shocks (originating from spillover effects and other types of contagion) are still evolving. Such linkages have seen a par-ticularly significant growth during the last decade (for example, Frank, González- Hermosillo, and Hesse 2008) and are therefore an important dimension to be captured by meaningful empirical analysis. This chapter focuses on the design of stress tests to capture spillover effects and demon-strates the potential impact based on a case study.

The first part of the chapter deals with the establishment of macro- financial scenarios that are explicitly informed by spillover effects. Scenario design for macroeconomic stress tests is typically based on an “indirect approach” (Jobst, Ong, and Schmieder 2013, Figure 12.1): (1) first, economic and financial variables are estimated conditional on a mac-roeconomic scenario; (2) in the second step, the trajectories of the economic and financial variables are translated into bank solvency and liquidity2 measures based on so- called satellite or auxiliary models. Three approaches have com-monly been used to predict economic and financial variables under stress (see Foglia 2009): first, a structural econometric

This chapter is based on IMF Working Paper 14/103 (Hesse, Salman, and Schmieder 2014). The authors thank Eugenio Cerruti, Martin Čihák, Amadou Sy, and participants at a Banque de France conference (September 2012) as well as at a meeting in May 2013 of the Liquidity Stress Testing Subgroup of the Basel Committee on Banking Supervision’s Research Task Force for helpful comments and suggestions.1 For work on stress testing at the IMF, for example, see Jobst, Ong, and Schmieder 2013.2 For liquidity stress tests, most tests have typically relied on the “direct” approach.


How to Capture Macro-Financial Spillover Effects in Stress Tests278

Specifically, the chapter infers from market data the mag-nitude of sovereign spread spillovers effects resulting from an increase in peripheral EU sovereign debt spreads, while con-trolling for changes in the market sentiment (that is, risk aversion) and macroeconomic factors. Using market data, the chapter seeks to capture point- in- time and dynamic time series’ effects, while recognizing the limitations of using market data, that is, that they might not necessarily “only” reflect underlying vulnerabilities and risks. The translation of sovereign spread spillovers into a loss of output is based on recent work at the IMF (Vitek and Bayoumi 2011).

Two approaches are used to capture the spillover effects in sovereign debt markets: panel regressions and a general-ized autoregressive conditional heteroskedasticity (GARCH) model. The panel regressions, which are used to establish an “average” impact of spillover effects during periods of stress on countries with advanced markets (AMs) and those with emerging markets (EMs), respectively, suggest that increas-ing sovereign risk in the euro periphery was a major driving force behind spillover effects. As expected, risk aversion, measured through changes in the VIX and high- yield spreads, is found to increase during periods of financial stress, exhibiting a nonlinear pattern. Country- specific mac-roeconomic factors also matter, but to a lesser degree, and their impact does not appear to change significantly under periods of stress.

GARCH models were run to obtain more granular spill-over effects, such as the country- specific comovements be-tween peripheral European GIIPS6 sovereign debt spreads and the corresponding spreads in the banks’ home countries (that is, the 25 most systemically important financial sys-tems, the “S25” sample) for specific points in time. The study reveals significant differences in terms of the spillovers across countries, with a higher impact observed for most core euro area countries (in particular during peak periods of the

The chapter aims to come up with a stress testing approach that captures spillover effects in detail. The solu-tion is an amended version of the indirect approach: the starting point is to establish a macroeconomic scenario, typi-cally not informed by potential spillover effects— at least not explicitly. In the second step, the potential marginal increase of stress due to spillover effects is estimated by translating the spillover effects into reduced output paths, that is, an adverse macroeconomic scenario.

The stylized design of macroeconomic stress tests (Figure 12.1) implicitly incorporates a quasi- feedback loop into the linear design of traditional stress tests— through a sensitivity- type approach.3 The approach could also include a test for interbank bank contagion, as shown in Figure 12.1. The chapter builds on previous IMF work to establish an ex-plicitly iterative process, that is, establish a scenario informed by initial spillover effects based on a structural econometric approach, compute the impact on banks’ solvency parame-ters, recompute the resulting spillover effects and feed them back to the structural model, and so on, until an equilibrium is reached.4 The approach presented here uses proxies for the “ultimate” impact of spillovers for different advanced and emerging economies conditional on the evolution of sover-eign spreads in the euro area periphery (that serves as the stress catalyst). Dynamic effects can also be captured via “di-rect” approaches, as done by Jobst and Gray 2013, for example, but renders the outcome a reduced- form type.5

3 Further information on macroeconomic scenarios used for financial sector assessment programs (FSAPs) can be found in Jobst, Ong, and Schmieder 2013.

4 At the IMF, such analyses were carried out by combining the work of Schmieder, Puhr, and Hasan (2011) and Vitek and Bayoumi (2011) as part of early- warning analysis and vulnerability exercises. It should be noted that running such an approach requires close cooperation be-tween staff running macroeconomic forecasts and staff simulating the impact of stress at the bank level (typically done by financial stability departments).

5 See also IMF 2011 and 2012a for further information on related work. 6 GIIPS refers to Greece, Ireland, Italy, Portugal, and Spain.

Macroeconomic scenario

Satellite models

Bank solvency and liquidityunder stress

Potential interbank contagion?

“Traditional” stress test

Macroeconomic scenario• Without contagion

• With contagion

Feedback typically not included

Source: Authors.

Figure 12.1 Stylized Design of Stress Tests


Heiko Hesse, Ferhan Salman, and Christian Schmieder 279

crisis) than for Scandinavian countries, Switzerland, the United Kingdom and most non- European countries. The findings also show a flight- to- quality element, that is, a neg-ative comovement of GIIPS spreads with German bunds and US Treasury bonds.

The second part of the chapter illustrates how the estab-lished spillover effects would feed through to banks based on a case study for 154 large international banks from the “S25” country sample. The impact of different degrees of spillover on banks’ solvency and liquidity positions is compared with baseline- type conditions (which correspond to realized stress scenarios in recent years, unlike in “normal” times). Stress at the bank level is simulated based on a recently developed IMF stress testing framework for liquidity (Schmieder and others 2012) and benefits from work on solvency (Schmie-der, Puhr, and Hasan 2011; Hardy and Schmieder 2013), which together allow running integrated solvency and li-quidity scenarios.7

The outcome suggests that spillover effects have a highly nonlinear impact on bank soundness, both in terms of li-quidity and solvency. It is thereby shown (once more) that the design of stress scenarios is a highly crucial element of stress testing, and is sensitive with respect to the outcome of stress tests.8 The magnitude of the impact on bank solvency and liquidity could serve as a benchmark for other studies, while recognizing that future spillover channels could be highly different, both in terms of direction and magnitude. In this sense, this study could help to identify potential sys-temic vulnerabilities ex ante, a role that stress tests have not necessarily played in the past for a number of reasons (see Borio, Drehmann, and Tsatsaronis 2012, for example).

The chapter is organized as follows. Section 2 investigates financial spillovers at the sovereign and bank level, based on panel regressions and a GARCH model framework. Section 3 provides a brief overview of the stress testing framework used to simulate the impact of spillover effects on bank liquidity and solvency. Section 4 shows the impact of different degrees of spillover based on a case study. Finally, Section 5 concludes and offers some avenues for future research.

2. FINANCIAL SPILLOVERS FROM THE EURO PERIPHERY TO THE REST OF THE WORLD

Panel Approach

Financial market linkages across economies have grown sig-nificantly in recent decades, which was felt strongly when the financial crisis started in 2008 with the failure of Lehman

Brothers, and later, when it continued to become a sovereign debt crisis, especially in the European periphery. AM finan-cial spillovers have been a dominant determinant of AM and EM financial soundness during the previous years.

Recent studies identified three important factors for spill-over effects (see, for example, Caceres and Unsal 2011): (1) a stress spillover catalyst— in this study AM sovereign debt yields; (2) risk aversion in global markets; and (3) country- specific risk factors.

This chapter seeks to establish benchmark parameters to simulate spillover effects at the bank level. Initially, a risk premium variable for the sample of 35 countries was con-structed.9 The risk premium is the spread between 10-year domestic treasuries to US Treasury bonds for non- European AM countries, to German bunds for AM countries in Eu-rope, and to the J.P. Morgan Emerging Markets Bond Index for the EM countries.10

Based on random effects’ panel regressions the sovereign spreads are regressed on three sets of peripheral spreads: aver-age spreads for (1) the European peripheral countries (GIIPS); (2) for the GIP (Greece, Ireland, Portugal); and (3) for IT- ES (Italy and Spain). Risk aversion is identified by two variables, high-yield spreads and the VIX. The former is the difference between yields to maturity of Moody’s Aaa- rated and Baa- rated US corporate bonds. The latter is the implied volatility for S&P 500 index options. Trade openness, liquidity (prox-ied by M2 to GDP and the level of reserves to GDP), inflation rates, GDP growth, the current account, the level of public debt and deficits- to- GDP ratios are used as macroeconomic control variables to capture country- specific cyclical effects.

The regressions are estimated for two time periods based on quarterly data: (1) 2006–12 and (2) 2008–12. The choice of the two sample periods is meant to capture the impact of the systemic stress.

The results (displayed in Appendix 12.1) present various model specifications considered useful to identify drivers of spillover stress and their actual impact, respectively. Us-ing the sovereign debt spreads of the 35 sample countries as the dependent variable, Appendix Table 12.1.1 shows the outcome for 2006–12 and Appendix Table 12.1.2 for 2008–12.

The results confirm previous studies in that all three fac-tors, that is, a catalyst, risk aversion, and country- specific factors, are actually important to explain financial stress (measured in terms of sovereign spreads), at least for the global financial crisis. Specifically:

• Increasing sovereign risk in the euro periphery was found to be a catalyst for spillover effects.

7 The frameworks were developed in the context of recent FSAPs and IMF technical assistance, extending the seminal work of Čihák 2007, and drawing upon work at the Austrian National Bank (OeNB 2013).

8 See also Taleb and others 2012 on how to test the sensitivity (that is, nonlinearity) of the outcome of stress tests.

9 The sample of countries includes Australia, Austria, Belgium, Brazil, Canada, China, Cyprus, Denmark, Finland, France, Germany, Greece, Hong Kong SAR, Hungary, India, Ireland, Italy, Japan, Korea, Luxem-bourg, Malta, Mexico, Netherlands, Norway, Poland, Portugal, Russia, Singapore, Slovenia, Spain, Sweden, Switzerland, Turkey, the United Kingdom, and the United States.

10 The panel regressions adjust for exchange rate changes.



framework is used for the estimation, which allows for hetero-skedasticity of the data and a time- varying correlation in the conditional variance. Specifically, the dynamic conditional correlation (DCC) specification by Engle 2002 is adopted, which provides a generalization of the constant conditional correlation model by Bollerslev 1990.13 The DCC GARCH models are estimated in first differences to account for the non-stationarity of the variables in the crisis period.

These econometric techniques allow us to analyze the daily comovement of the GIIPS spreads and the sovereign bond spreads of the sample of AMs and EMs. The GIIPS spreads are included in the model as a conditioning variable, as is the VIX. The methodology is therefore closely aligned to the one of the panel regression and further explained in Appendix 12.2.

The sample period chosen was daily data from 2007 to the end August 2012, with a view to cover the full crisis pe-riod. As before, for the European AMs the risk premium of 10-year instruments was measured as the difference between the average GIIPS spread as well as those of the domestic treasuries to German bunds. For the non- European coun-tries, the spread to the 10-year US Treasury bonds is calcu-lated and for EM countries the J.P. Morgan Emerging Markets Bond Index global spread and the HSBC Asian US dollar spread for Asian countries are used.

As expected, our findings suggest that the spread between GIIPS to German bunds exhibits a higher degree of comove-ment with the risk premia for European countries than non- European countries (Figures 12.2–12.5). In particular, implied DCC GARCH correlations with the GIIPS spread were as high as 0.7–0.8 for Austria, Belgium, France, and the Netherlands during episodes of systemic stress (Figure 12.2, panels 1 and 2). In contrast, the GIIPS comovement with the United Kingdom spread to German bunds is relatively low and oscillates between 0 and 0.2, while the model- implied correla-tion with the Swiss spreads reaches a maximum of 0.4 (Fig-ure 12.2, panel 3). The results also show that the spreads of the Scandinavian countries, namely Denmark, Norway, Sweden, and Finland (though with higher average levels),14 on average exhibit a lower comovement with the GIIPS spread than do their continental European peers (Figure 12.2, panel 4). The outcome also suggests a constant level of stress, with some eas-ing toward the end of the observation period, a finding which also applies to the non- European sovereigns.

Comovements of the GIIPS spread with Australian and Canadian spreads (relative to US Treasury bonds) are rather low, with implied correlations up to 0.2 (Figure 12.3, panel 1). Looking at the Asian countries Hong Kong, Japan, and Sin-gapore shows a somewhat higher correlation with the GIIPS

• The global perception of risk magnifies stress condi-tions as do expected future interest rates.

• Country- specific macroeconomic factors also matter, but to a lesser degree.

• While the impact11 of country- specific factors does not appear to change significantly under stress, the impact of the former two factors is higher during 2008–12, that is, in the period covering only the cri-sis years (compared to the full sample period).

For the longer sample period (that is, 2006–12) a 1- percentage-point change in euro periphery sovereign spreads (that is, GIIPS and GIP) translates into a 0.2- to 0.3- percentage-point change of sovereign debt spreads in the 35 sample countries (Appendix Table 12.1.1). Global risk aver-sion (measured by changes in high- yield spreads) has an even higher impact— a 1-percentage-point change in high- yield spreads translates into about a 0.6-percentage-point change in sovereign spreads. As global risk aversion and high- yield spreads are highly correlated during episodes of stress, the joint impact on the peripheral spreads is exacerbated, which is illustrated in a comparison of the coefficients in Appendix Tables 12.1.1 and 12.1.2. The transmission of risk premium shocks from Italy and Spain to the countries in the sample is more pronounced than for the GIPs. Depending on the model specification, the availability of domestic liquidity and trade openness also contribute to some degree to spillovers.12

The outcome for the crisis period only (covering the years from 2008 to 2012, Appendix Table 12.1.2) indicates that the coefficients for all three major drivers, that is, European pe-riphery shocks and global risk aversion, as well as the slope of the US yield curve, are higher than for the period including precrisis years (Appendix Table 12.1.1). A 1 percent shock to euro periphery spreads translates into a 0.5- percentage-point increase in the risk premium of the 35 sample countries if the shock originates in the GIPs and a 1-percentage-point increase in spreads if it originates in Italy and Spain. Hence, it seems that the size of the peripheral European country determines the size of spillovers, as expected. Moreover, global risk- aversion shocks also translate almost one- to- one into spreads.

DCC GARCH Approach

The panel regression approach provided the average spillover effect on countries’ sovereign spreads. Later in this chapter, the previous work is complemented by estimating country- specific daily comovements, in order to differentiate more between countries, and to come up with the range of the potential spill-over impact observed over time. A multivariate GARCH

11 Measured in terms of the R- squared and the actual coefficients.12 For a robustness check, a separate set of regressions were run to estimate

the impact of expectations of higher interest rates, represented by the slope of the US Treasury yield curve on the global risk premium. Re-sults indicate that a steepening of the curve implies higher costs of bor-rowing for the periphery countries.

13 Given the high volatility movements during the global financial crisis, the assumption of constant conditional correlation among the variables in the constant conditional correlation model is not very realistic, espe-cially in times of stress where correlations can rapidly change. Therefore, the DCC model is a better choice, since correlations are time- varying.

14 Finland is the only euro- area country within the sample, which seems to explain the higher level of correlations.



Since the onset of sovereign debt crisis by 2009, the aver-age GIIPS interest rates exhibit a negative correlation with both the German bund and US Treasury bond interest rates (Figure 12.5). Since 2009, the implied correlation has turned negative for both countries, with lows at -0.4 (United States) and -0.6 (Germany), indicating a sudden flight to safety, in line with other recent studies (IMF 2011, for example).

spread of up to 0.3 and with one jump to 0.4 (Figure 12.3, panel 2). In terms of EM countries, results suggest that Chi-na’s comovement with the GIIPS spread is rather subdued compared to the other EM countries: Brazil, Mexico, Russia, and Turkey (Figure 12.4). Out of this EM sample, Turkey has the highest implied correlation with the GIIPS during epi-sodes of system stress at up to 0.6.

GIIPS-BelgiumGIIPS-France GIIPS-Austria

GIIPS-Netherlands

GIIPS-SwitzerlandGIIPS-UK

GIIPS-Denmark GIIPS-FinlandGIIPS-Norway GIIPS-Sweden

–0.2

0.9

0

0.6

0.8

0.4

0.2

–0.1

0.5

0.7

0.3

0.1

1. Belgium and France

–0.05

0.5

0.05

0.35

0.45

0.25

0.15

0

0.3

0.4

0.2

0.1

3. Switzerland and the United Kingdom

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

–0.1

0.8

0

0.6

0.4

0.2

0.5

0.7

0.3

0.1

2. Austria and Netherlands

0

0.6

0.4

0.2

0.3

0.5

0.1

4. Austria and Canada

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

Sources: Bloomberg; and authors’ calculations.Note: GARCH = generalized autoregressive conditional heteroskedasticity; GIIPS = Greece, Ireland, Italy, Portugal, and Spain.

Figure 12.2 Estimated GARCH Correlations GIIPS with European Countries

GIIPS-Hong Kong SARGIIPS-Japan GIIPS-Singapore

GIIPS-AustraliaGIIPS-Canada

–0.2

0.5

0

0.2

0.4

–0.1

0.1

0.3

1. Australia and Canada

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

–0.05

0.25

0

0.15

0.05

0.1

0.2

2. Hong Kong SAR, Japan, and Singapore

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

Sources: Bloomberg; and authors’ calculations.Note: GARCH = generalized autoregressive conditional heteroskedasticity; GIIPS = Greece, Ireland, Italy, Portugal, and Spain; Hong Kong SAR = Hong Kong Special Administrative Region.

Figure 12.3 Estimated GARCH Correlations GIIPS with Non- European Countries



work presented in Schmieder and others 2012 was developed in the context of recent FSAPs15 and IMF technical assis-tance, extending the seminal work of Čihák 2007 and

3. LIQUIDITY AND SOLVENCY STRESS TESTINGThe area of stress testing has seen a number of advances dur-ing recent years. This study uses a recently developed IMF liquidity stress testing framework to run integrated solvency and liquidity stress tests. The liquidity stress testing frame-

15 Examples include Chile, Germany, India, Spain, Turkey, and the United Kingdom.

GIIPS-Brazil GIIPS-ChinaGIIPS-Mexico

GIIPS-Poland GIIPS-RussiaGIIPS-Turkey

GIIPS-India GIIPS-Korea

–0.1

0.6

0

0.4

0.2

0.5

0.3

0.1

1. Brazil, China, and Mexico

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

–0.05–0.1

–0.15

0.3

0.05

0.25

0.15

0

0.2

0.1

3. India and Korea

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

–0.1–0.2

0.7

0

0.4

0.2

0.50.6

0.3

0.1

2. Poland, Russia, and Turkey

Jan. 2,11

Jan. 2,12

Jan. 2,10

Jan. 2,09

Jan. 2,08

Jan. 2,2007

Sources: Bloomberg; and authors’ calculations.Note: EM = emerging market; GARCH = generalized autoregressive conditional heteroskedasticity; GIIPS = Greece, Ireland, Italy, Portugal, and Spain.

Figure 12.4 Estimated GARCH Correlations GIIPS with EM Countries and Korea

GIIPS-GermanyGIIPS-United States

–0.2

–0.4

–0.6

–0.8

1

0

0.6

0.2

0.4

0.8

Jan. 2, 11 Jan. 2, 12Jan. 2, 10Jan. 2, 09Jan. 2, 08Jan. 2, 2007

Sources: Bloomberg; and authors’ calculations.Note: Unlike the other GARCH models, the average GIIPS interest rates are taken and not the GIIPS spread to German bunds. GARCH = generalized autoregressive conditional heteroskedasticity; GIIPS = Greece, Ireland, Italy, Portugal, and Spain.

Figure 12.5 Estimated GARCH Correlations GIIPS with Germany and the United States



Nevertheless, while this chapter attempts to condense a wealth of information and assumptions to establish integrated scenarios this should not, in any sense, give a false sense of precision. Instead, it is recommended that a whole range of scenarios be run, which can build upon the ones established in the study, with varying degrees of severity. Reverse stress tests can be also included.18 This is an important way forward to obtain a better understanding of key solvency and liquidity risks faced by banks, and to gain a more comprehensive view on their respective risk tolerances.

Liquidity Stress Testing Approach

An implied cash- flow approach is applied to simulate the im-pact of a bank- run- type stress scenario (Appendix 12.3). The banks’ liabilities are broken down into demand and term de-posits, short- term wholesale funding (including bank and secured funding), derivative funding as well as long- term funding such as senior debt or subordinated debt. On the

drawing upon work at the Austrian National Bank.16 An overview of recent academic and policy research on integrat-ing liquidity and solvency stress testing is given in Box 12.1.

In this study, the focus is on scenario design, namely building integrated scenarios for solvency and liquidity risks that take into account spillover effects and feedback loops.17 The central question becomes how the findings established in Section 2 can be used to inform bank- level stress tests.

16 It is complemented by a previously developed solvency stress testing tool by Schmieder, Puhr, and Hasan 2011. While developing the solvency and liquidity stress testing frameworks, four key facts were accounted for, which constitute key challenges of contemporaneous financial stabil-ity analysis: (1) the availability of data varies widely, and lack of data is common; (2) both solvency and liquidity risk have various dimensions, which requires multidimensional analysis, thereby integrating risks; (3) designing and calibrating scenarios is challenging, even more so for li-quidity risk than for solvency risk (mainly as liquidity crises are relatively rare and originate from different sources); and (4) communication of stress test results is a key integral part of the exercise. The answer to these multiple dimensions is Excel- based balance- sheet- type frameworks.

17 The exercise thereby reflects key principles for liquidity stress testing put forward by the Basel Committee in the aftermath of the first wave of shocks following the default of Lehman Brothers (BCBS 2008).

18 The work by Taleb and others 2012 and Hardy and Schmieder 2013, for example, could be useful to consider in this context.

Box 12.1. Integrating Liquidity, Solvency Risks, and Bank Reactions in Stress Tests

Banks have numerous and overlapping ways to react to credit and funding shocks. High- quality capital and profits are usually the first line of defense, and retained earnings can help buffer banks’ capital levels. In terms of liquidity, banks have an inherent counterbalanc-ing capacity to generate liquid assets by using high- quality eligible securities as collateral to generate market funding, or, if interbank markets freeze entirely, central bank funding. As seen post- Lehman, fire sales of securities can also be an option to generate liquidity, but at a considerable cost in an environment of sharply declining asset prices. Deleveraging, especially targeted at assets with higher risk weights, is also a way to raise capital adequacy ratios by reducing risk- weighted assets. In practice, banks have been using a combination of these, as well as other hybrid measures, ranging from debt- to- equity conversions, to issuance of convertible bonds, to optimizing risk- weighted assets, to react to shocks.

Incorporating banks’ reactions to shocks is a critical component for the design of informative stress tests, especially over lon-ger time horizons. This, however, requires modeling solvency and liquidity shocks in a coherent manner because first, when banks react to financial stress, the source of the shock (solvency or liquidity) is not always clear; and second, the measures banks take in reaction to these shocks have both capital and liquidity aspects that are not easy to disentangle.

Recently, a number of analytical approaches have attempted to integrate solvency and liquidity more systematically. • Empirical work includes Van den End 20081 at the Dutch Central Bank and Wong and Hui 2009 from the Hong Kong Monetary

Authority,2 for example. Barnhill and Schumacher (2011) developed a more general empirical model, incorporating the previous two approaches that attempt to be more comprehensive in terms of the source of the solvency shocks and compute the longer term impact of funding shocks.

• Schmieder and others (2012) provide an Excel- based framework that allows running liquidity tests informed by banks’ solvency con-ditions, and to simulate the increase in funding costs resulting from a change in solvency.

• An integrated approach to model funding liquidity risks and solvency risk is the Risk Assessment Model for Systemic Institutions de-veloped by the Bank of England (Aikman and others 2009). The framework simulates banks’ liquidity positions conditional on their capitalization under stress and other relevant dimensions such as a decrease in confidence among market participants under stress. A recent attempt by the Austrian National Bank to come up with an integrated framework and to overcome operational challenges identified with previous work on integrated models, the Applied Risk, Network and Impact Assessment Engine, should also be men-tioned (OeNB 2013).

For an overview of liquidity stress tests, including the link to solvency, see also BCBS 2013. IMF 2013 examines the European Banking Authority stress tests.

Source: Oura and Schumacher 2012.

1 Van den End (2008) developed a stress testing model that tries to endogenize market and funding liquidity risk by including feedback effects that capture both behavioral and reputational effects. A number of central banks and bank supervisors have been successfully using the Monte Carlo framework of Van den End 2008.

2 The authors sought to explicitly capture the link between default risk and deposit outflows. Their framework allows simulating the impact of mark- to- market losses on banks’ solvency position leading to deposit outflows; asset fire sales by banks are evaporating and contingent liquidity risk sharply increases.



nomic scenario, and is used as a sensitivity analysis. The translation of the spillover effects into the revised mac-roeconomic trajectories is based on recent IMF work.

3. Soundness of banks: The scenario is translated into bank- level stress parameters to simulate both the banks’ solvency and liquidity positions, drawing on work by Hardy and Schmieder 2013 and Schmieder and others 2012, respectively.

Bank- level data from BankFocus (from the end of June 2012) was used for large systematically important banks. In total, the sample includes 154 large banks from the following 26 countries: Austria, Australia, Belgium, Bra-zil, Canada, Switzerland, China, Germany, Denmark, Finland, France, United Kingdom, Hong Kong Special Administrative Region, India, Japan, Korea, Luxembourg, Mexico, Nether-lands, Norway, Poland, Russia, Sweden, Singapore, Turkey, and the United States.

The sample comprises almost the full European Banking Authority sample for the European banks (except for the banks in the GIIPS countries) and includes the largest banks in the non- European countries. In total, it captures $84 tril-lion of bank assets (that is, about 50 percent of the assets held by banks worldwide), $39 trillion nonbank deposits, and about $7 trillion of government securities held by banks.

Scenarios

Four different scenarios are referred to: The April 2012 WEO baseline scenario for 2013–14 (Scenario 1); and three spillover scenarios (referred to as Scenarios 2.X) conditional on Scenario 1—scenarios that banks could potentially face in case increas-ing degrees of spillovers affect the general growth trend.

Specifically, Scenario 1 is adjusted for an increase of GIIPS spreads by 100 (Scenario 2a), 200 (Scenario 2b), and 300 (Scenario 2c) basis points, respectively. The study fur-ther distinguishes between the spillover impact observed during periods of substantial financial stress (using the panel regression for 2008–12 and the GARCH model for 2010–12) and during periods of less significant stress (using the panel regression for 2006–12 and the GARCH model

asset side, a range of potentially liquid asset positions is in-cluded, such as cash, government, trading, and investment (both available- for- sale and held- to- maturity) securities, loans and advances to banks, reverse repos, and cash collat-eral. Given European periphery banks’ increasing collateral use of pools of loans (such as covered bonds) for liquidity, a crude definition of banks’ loan level as a portion of their to-tal assets is also included.

Solvency Stress Testing Approach

Rules of thumb for solvency stress testing are used, as pro-posed by Hardy and Schmieder 2013, and thereby a simplified solvency test.19 Credit losses, banks’ preimpairment income, and the trajectories of RWAs for a two- year horizon were simulated based on the GDP trajectories, with and without spillover effects. The capital shortfall was measured against a Tier 1 capital ratio (Tier 1 capital/RWAs) of 6 percent, below which a bank is considered undercapitalized.20

4. INTEGRATION OF THE FINANCIAL SPILLOVER ANALYSIS WITH THE STRESS TESTING APPROACHThis integrated approach to simulate stress at the bank level is illustratively shown in Figure 12.6:

1. Scenario design: GDP trajectories of a specific mac-roeconomic scenario were used, the World Economic Outlook (WEO) baseline scenario for 2013–14 as of April 2012, and the spillover stress component was added.

2. Spillover analysis: The outcome of the spillover analysis (see previous sections of this chapter), measured through a widening of sovereign spreads, worsens the macroeco-

Scenario(for example,

WEO)

GDP trajectory,adjusted for

spillovereffects

IMF spillover analysis panel/GARCH

Bank solvency parameters:• Credit losses• Security P/L impact• Preimpairment income

Bank liquidity parameters• Haircuts (market liquidity)• Outflow of funding (funding liquidity)

Overall soundness of bankTranslation into bank-level stress scenario:

Solvency: Hardy and Schmieder (2013)Liquidity: Schmieder and others (2012)

Source: Authors.Note: GARCH = generalized autoregressive conditional heteroskedasticity; P/L = profit/loss; WEO = World Economic Outlook.

Figure 12.6 Overview of the Concept to Simulate Stress at the Bank Level

19 However, it should be noted that the evidence is based on a comprehen-sive set of data from 16,000 banks during the last 15 years (as available).

20 Please note that this specific choice is meant for illustration only—through a similar level as used for the European stress tests conducted in 2010 and 2011, for example.



impairment income for 2013 and 2014.21 For a stylized bank with loss impairment rates of 0.5 percent and a preimpairment return on capital of 10 percent in 2012,22 loan impairment rates are simulated to decrease slightly under the baseline sce-nario and mild spillover conditions, while they would increase (nonlinearly) under increasing levels of spillover stress. The same pattern holds for preimpairment income. This input is used to simulate the bank’s capital, risk- weighted, and capital ratio. Again, the same pattern holds, with a decrease of the styl-ized banks’ assets,23 capital ratio to 7.5 percent under the most severe scenario, which is above the hurdle rate in terms of Tier 1 capital to pass the stress test (6 percent).

The outcome of this solvency stress test applied to the 154 banks presented in Figure 12.7 shows that the large interna-tional banks would be in a position to digest the baseline scenario plus some level of spillover stress, while additional stress in the euro area periphery results would have a highly nonlinear impact on potential capital needs. The nonlinear-ity results from two factors: (1) the nonlinearity in the satel-lite models for loan impairment rates and preimpairment income, and (2) the effect of the kick- in of capital needs for banks that fall below the hurdle rate.

Impact on Bank Liquidity

For the liquidity stress test, the study simulates the impact of stress on both banks’ market liquidity (that is, their ability to fire sale assets) and funding liquidity (that is, the potential outflow of funding).24 Again, it was assumed that the bank is affected by the shock in its home country.25

for 2008–12), that is, refer to a total of six spillover scenarios (2a/1, 2a/2, 2b/1, 2b/2, 2c/1, 2c/2).

For the banks’ solvency, their Tier 1 capital ratios are stimulated by end- 2014, based on the evolution of the main solvency dimensions (banks’ income and losses). For liquid-ity, the impact is determined of a worst- case idiosyncratic shock to the bank’s liquidity profile on top of the impact on liquidity resulting from the macroeconomic/spillover sce-narios. Illustrative examples are provided in Appendix 12.4 (solvency) and Appendix 12.5 (liquidity).

Impact on Bank Solvency

As outlined earlier in this chapter, the study uses the outcome of the IMF’s 2012 Spillover Report (IMF 2012b), which simu-lates the impact of a 300-basis-point increase in peripheral countries’ spreads (including a lower yield increase for core countries) on European countries’ GDP paths based on the IMF G35 model (drawing upon Vitek and Bayoumi 2011).

Appendix 12.4 provides an illustrative example for a stylized Austrian bank. In the first step, the increase of Austrian sover-eign debt spreads is simulated, using the evidence established in Section 2. A 100-basis-point shock of GIIPS spreads (Scenario 2a) would thereby result in an increase of Austrian spreads by 24 basis points for less significant spillover stress (Scenario 2a/1) and 50 basis points (2a/2) for more substantial spillover stress. Measured relative to the April 2012 WEO baseline scenario for Austria, suggesting real GDP growth rates of 1.8 percent (2013) and 2.2 percent (2014), spillover analysis carried out at the IMF (2012b) would predict a drop of real GDP growth by about 0.45 percentage point for Scenario 2a/1 (less significant spill-over stress), whereby the GDP trajectory becomes 1.4 percent (2013) and 1.8 percent (2014). For a period with more signifi-cant spillover (Scenario 2a/2), the impact is about twice (0.9 per-centage point), whereby the GDP trajectory is 0.9 percent (2013) and 1.3 percent (2014). For a 200-basis-point shock (Scenario 2b), growth drops by 1.7 percentage point and for 300 basis points (Scenario 2c) by 2.6 percentage points (per year) under substantial spillover conditions (Appendix 12.4).

The satellite models by Hardy and Schmieder (2013) are then used to determine banks’ loan impairment levels and pre-

21 For simplification, it was assumed that banks are affected according to their domestic scenarios, that is, that their businesses are predominantly based in their home countries.

22 In a few cases, the latest available figures were from 2011.23 The risk- weighted assets are simulated based on work by Schmieder,

Puhr, and Hasan (2011), assuming point- in- time credit risk parameters.24 Unlike for the solvency scenario, the study does not simulate stress for a

specific point in time; rather, the simulated stress conditions reflect a worst- case situation resulting from the general macroeconomic condi-tions as well as an idiosyncratic shock to the bank conditional.

25 In other words, it is assumed that all of its assets are based in the home country, which is a crude simplification.

Less substantial spillover stressMore substantial spillover stress

0

25

15

5

10

20

Baseline (Sc 1) Sc 2a (Shock 100bps) Sc 2b (Shock 200bps) Sc 2c (Shock 300bps)

Capital needs (Tier 1, billions of US dollars) for different scenarios

Source: Authors.Note: bps = basis points; Sc = Scenario.

Figure 12.7 Outcome of Solvency Stress Tests



bank sample for Scenario 2c/2 (300-basis-point spread shock, significant spillover stress) and close to 6 percent for Scenario 2b/2 (200-basis-point spread shock), compared to 0.3 percent and 1 percent if measured against total assets.26

5. CONCLUSIONThis study attempts to address the challenge faced by current financial stability analysis, namely to capture spillover ef-fects and other types of contagion that ultimately determine macro- financial stress at the bank level.

By integrating recent IMF work on financial spillover analysis and stress testing, the study uses a novel framework that allows shedding some light on the potential impact of spillover effects on bank- level solvency and liquidity. Never-theless, it is recognized that significant additional effort and evidence are needed to make the modeling of dynamic macro- financial linkages more robust, not least due the many po-tential channels of spillover and contagion, the fact that the use of crude data available for stress tests is subject to uncer-tainty, and other factors that contribute to uncertainty (such as mixed evidence for the use of market data).

The outcome of the stress tests suggests that spillover ef-fects observed for the sovereign debt markets in recent years have a highly nonlinear impact on bank soundness, both in terms of liquidity and solvency. This implies (once more) that the design of stress scenarios is a crucial element of stress testing, and is very sensitive with respect to the out-come of stress tests. The approach used in this chapter is meant to be a menu for future analyses of the impact of po-tential spillovers. Sensitivity analysis and reverse stress tests appear to be an important complement in this context.

The link between the level of stress and bank liquidity is established based on empirical work of Schmieder and others 2012. The study links the GDP trajectories implied by the changes of sovereign spreads to funding shocks experi-enced by the most affected banks during the Lehman crisis. In other words, the study simulates highly adverse idiosyncratic liquidity shocks conditional upon macroeconomic conditions.

In line with (very limited) empirical evidence, it was expected that the relationship between the shock and the potential adverse impact on the bank level would be highly nonlinear (as implied by the scenarios in Appendix 12.3, and in addition to the nonlinearity for the banks hitting the hurdle rate, as for capital). Under a worst case scenario, banks would experience a shock equal to a “Lehman Brothers–type” scenario, the “severe stress scenario” in Appendix 12.3. This shock level represents how the stress at the time of the Lehman Brothers event affected the banks that were most severely hit, that is, overlays a market shock with an idiosyn-cratic liquidity shock. The stress level relative to the one ex-perienced by banks at the time of the Lehman Brothers crisis is established via the cumulative GDP trajectory under stress compared to the long-term average. For the stylized example presented in Appendix 12.5, the stress level is at 0.65, that is, the benchmark funding stress parameters (for the “severe stress scenario”) in Appendix 12.3 have to be multiplied by 0.65. The funding available for the specific banks under the European Central Bank’s Long Term Refinancing Opera-tions is inferred from country- level data and used as a cush-ion for the relevant European banks.

Figure 12.8 shows the outcome of this liquidity stress test. Under the baseline scenario all banks have sufficient li-quidity, as expected. Adding spillover stress triggers a non-linear increase of liquidity needs (which occur in case the liquidity needs exceed the available liquidity generated via fire sales), and more substantial spillover stress makes the stress highly nonlinear. Measured against Tier 1 capital rather than total assets, the substantial spillover stress leads to a maximum liquidity shortfall of 20 percent for the entire

Less substantial spillover stressSubstantial spillover stress

0.0%

1.2%

0.6%

0.2%

0.4%

0.8%

1.0%

Baseline (Sc 1) Sc 2a (Shock 100bps) Sc 2b (Shock 200bps) Sc 2c (Shock 300bps)

Liquidity Needs as a Percentage of Total Assets

Source: Authors.Note: bps = basis points; Sc = Scenario.

Figure 12.8 Outcome of Liquidity Tests in Terms of Assets

26 The study did not explicitly model a central bank response as the Lender of Last Resort to mitigate the estimated liquidity shortfall. In reality and as seen during the crisis period, central banks would provide large liquidity support to solvent banks, subject to an appropriate haircut.


Appendix 12.1.Outcome of Panel Regressions

Assessing Spillover Risks


Panel Regressions, 2006:Q1–2012:Q2 (Dependent variable: sovereign spreads of 35 sample countries)(Quarterly data)Explanatory Variables1 (1) (2) (3) (4) (5) (6)GIIPS spread 0.237***

(0.045)0.244***

(0.047)GIP spread 0.288***

(0.046)0.289***

(0.047)Italy/Spain spread 0.611***

(0.09)0.653***

(0.094) High- yield spread 0.666***

(0.242)0.621***

(0.229)0.357

(0.30)VIX 0.348

(0.238)0.342

(0.229)-0.070(0.291)

Openness 0.015(0.017)

0.015(0.017)

0.031*(0.016)

0.030*(0.016)

0.025(0.020)

0.025(0.021)

M2/GDP 0.080***(0.017)

0.078***(0.017)

0.061***(0.016)

0.060***(0.016)

0.053***(0.020)

0.051**(0.020)

Constant 0.297**(0.131)

-0.632(0.744)

0.256*(0.136)

-0.660(0.718)

0.700***(0.166)

0.997(0.912)

R2 (within) 0.77 0.70 0.79 0.73 0.79 0.78Observations 415 415 435 435 454 454T 25 25 23 23 26 26

Source: Authors.Note: Standard errors in parentheses. GIIPS = Greece, Ireland, Italy, Portugal, and Spain; GIP = Greece, Ireland, and Portugal; M2 = M2 money supply; T = Number of quarters covered by the regressions; VIX = CBOE Volatility Index.*p < 0.1; **p < 0.05; ***p < 0.01. 1 Right- hand- side variables are in logs.




Panel Regressions, 2008:Q1–2012:Q2 (Dependent variable: sovereign spreads of 35 sample countries) (Quarterly data)

Explanatory Variables1 (1) (2) (3) (4) (5) (6)GIIPS spread 0.492***

(0.105)0.463***

(0.106)GIP spread 0.511***

(0.090)0.479***

(0.090)Italy/Spain spread 1.002***

(0.173)0.998***

(0.175)High-yield spread 1.042***

(0.299)1.033***

(0.279)0.735**

(0.366)VIX 0.823**

(0.322)0.813***

(0.301)0.517

(0.397)Openness 0.018

(0.021)0.017

(0.021)0.034*

(0.019)0.033*

(0.019)0.033

(0.027)0.032

(0.027)M2/GDP 0.078***

(0.020)0.075***

(0.020)0.057***

(0.018)0.056***

(0.018)0.045*

(0.025)0.043*

(0.025)Constant -0.133

(0.222)-2.418**(1.084)

-0.216(0.222)

-2.459**(1.022)

0.308(0.246)

-1.117(1.307)

R2 (within) 0.93 0.78 0.91 0.78 0.91 0.85Observations 321 321 357 357 341 341T 18 18 18 18 18 18

Source: Authors.Note: Standard errors in parentheses. GIIPS = Greece, Ireland, Italy, Portugal, and Spain; GIP = Greece, Ireland, and Portugal; M2 = M2 money supply; T = Number of quarters covered by the regressions; VIX = CBOE Volatility Index.*p < 0.1; **p < 0.05; ***p < 0.01.1Right-hand-side variables are in logs.


Main Explanatory VariablesFactor Variable DescriptionSovereign risk GIIPS spread Average of euro periphery sovereign spreads

to German bundsGIP spread Average of Greece, Ireland, and Portugal

sovereign spreads to German bundsItaly/Spain spread

(IS spread)Average of Italy and Spain sovereign spreads

to German bundsRisk aversion High-yield spread Difference between yields to maturity of

AAA-rated and BAA-rated corporate US bonds

VIX Implied volatility of S&P 500 index optionsMacroeconomic

environmentOpenness Sum of imports and exports to GDP ratioM2/GDP Broad money to GDP ratio

Source: Authors.Note: GIIPS = Greece, Ireland, Italy, Portugal, and Spain; GIP = Greece, Ireland, and Portugal; Italy-Spain; M2 = M2 money supply; S&P = Standard & Poor’s; VIX = CBOE Volatility Index.


Appendix 12.2.Outline of the DCC GARCH Method

The dynamic conditional correlation (DCC) model is estimated in a three- stage procedure. Let rt denote an n × 1 vector of asset returns, exhibiting a mean of zero and the following time- varying covariance:

1 (0, )r

t t N Dt Rt Dtℑ − ∼ (Appendix 12.2.1)

{ }=where .Dt diag hit Here, Rt is made up from the time-dependent correlations, and Dt is defined as a diagonal matrix comprised of the standard deviations implied by the estimation of univariate generalized autoregressive conditional heteroskedasticity (GARCH) models, which are computed separately, whereby the ith element is denoted as hit . In other words, in this first stage of the DCC esti-mation, univariate GARCH models are fit for each of the five variables in the specification. In the second stage, the intercept parameters are obtained from the transformed asset returns, and in the third stage, the coefficients governing the dynamics of the conditional correlations are estimated. Overall, the DCC model is characterized by the following set of equations (see En-gle 2002 for details):

ω κ λε

ε ε

ε ε[ ]

= + ′ +== ′ − − + ′ +=

= ′

− − −

−

− − −− −

D diag diag r r diag i D

D r

Q S u A B A B Q

R diag Qt Qt diag QtS E

t i i t t t

t t t

t t t t

t

t t

� �

� � �

{ } { } { }

( )

{ } { }

.

21 1 1

2

1

1 1 1

1 1

(Appendix 12.2.2)

Here, S is defined as the unconditional correlation matrix of the residuals εt of the asset returns rt. As defined above, Rt is the time- varying correlation matrix and is a function of Qt, which is the covariance matrix. In the matrix Qt,ι is a vector of ones, A and B are square and symmetrical, and � is the Hadamard product. Finally, λi is a weight parameter with the contributions of Dt 1

2− declining over time, while κi is the parameter associated with the squared lagged asset returns. The estimation frame-

work is the same as in Frank, Gonzalez- Hermosillo, and Hesse 2008 and Frank and Hesse 2009.


Appendix 12.3.Benchmark Liquidity Stress Scenarios


Benchmark Liquidity Stress ScenariosScenario Moderate Stress

ScenarioMedium Stress Scenario

Severe Stress Scenario

Very Severe Stress Scenario

Severity (x times Lehman1) 0.25 0.5 1 2Liquidity OutflowsCustomer DepositsCustomer deposits (term) 2.5 percent 5 percent 10 percent 20 percentCustomer deposits (demand) 5 percent 10 percent 20 percent 40 percentWholesale FundingShort term (secured) 5 percent 10 percent 20 percent 40 percentShort term (unsecured) 25 percent 50 percent 100 percent 100 percentContingent liabilities 0 percent need funding 5 percent need

funding10 percent need

funding20 percent need

fundingLiquidity InflowsHaircut for cash 0 percent 0 percent 0 percent 0 percentHaircut for government

securities2

1 percent 2 percent 5 percent 10 percent

Haircut for trading assets3 3 percent 6 percent 30 percent 100 percentProxies, specific assets Equities: 3;

Bonds: 3 Equities: 4–6;

Bonds: 3–8Equity: 10–15;

Bonds (only LCR eligible ones): 5–10

Not liquid

Haircut for other securities 10 percent 30 percent 75 percent 100 percentProxies, specific assets Equities: 10;

Bonds: 10 Equities: 25;

Bonds: 20 (some not liquid)

Equity: 30; Bonds (only LCR eligible ones): 20–30

Not liquid

Percent of liquid assets encumbered4

10 percent (or actual figure)

20 percent (or actual figure plus 10 ppt)

30 percent (or actual figures plus 20 ppt)

40 percent (or actual figures plus 30 ppt)

Source: Schmieder and others 2012.Note: LCR = liquidity coverage ratio; ppt = percentage points.1The Lehman-type scenario would correspond to a scenario encountered by banks that were hit severely during the 30-day period after the Lehman col-lapse, that is, a stress situation within a stress period rather than an average. The scenario has been put together based on expert judgment, using evi-dence as available.2The haircut highly depends on the specific features of the government debt held (rating, maturity, market depth) and can be higher or lower. The figures displayed herein are meant for high-quality investment-grade bonds, taking into account recent market conditions. The same applies for the remainder of the liquid assets. For the securities in the trading book, it is assumed that they are liquidated earlier, resulting in lower haircuts.3A haircut of 100 percent means that the asset is illiquid, that is, the market has closed.4The figures account for a downgrade of the bank, which triggers margin calls, and higher collateral requirements more generally. Please note that the unencumbered portion applies to a gradually narrower definition of liquid assets.


Appendix 12.4.Illustrative Example for the

Solvency Test

Table 12.4.1 provides an illustrative example for a hypothetical bank in Austria.


Solvency: Illustrative Example for a Hypothetical Austrian BankStep 1.1. Spillover Impact in Sovereign Debt Markets Observed for Austria

Scenario (increase of GIIPS sovereign debt spreads by . . .)

Impact on Austria (Average for 2006–12)

Impact on Austria during Peak Spillover Stress (2008–12)

Increase of spreads (bps)

Source Increase of spreads (bps)

Source

100 bps (2a) 24.41

(=24*1.017)Appendix Table 12.1.1,

spec (22), Figure 12.2, panel 2

49.83 (=49*1.017)

Appendix Table 12.1.2, spec (1), Figure 12.2, panel 2

200 bps (2b) 48.8 Same, linear increase assumed

99.6 Same, linear increase assumed

300 bps (2c) 73.2Same, linear increase

assumed 149.4Same, linear increase

assumed

Step 1.2. GDP Trajectory for Austria, Adjusted for the Impact of Spillovers (GDP elasticity of widening of spreads for Austria estimated for two-year period from 2013–14: 3.5 [based on IMF 2012b])4

Trajectory based on evidence for 2006–12 (less significant spillovers)

Scenario 2012 2013 2014 Cumulative deviation of output from 2012 real GDP growth

level (2013–14), pptsBaseline (1) 0.9 1.8 2.2 2.2(2a/1) 0.9 1.4

(=1.8–0.5*3.5*0.244)1.8

(=2.2–0.5*3.5*0.244)1.3

(2b/1) 0.9 0.9(=1.8–0.5*3.5*0.488)

1.3(=2.2–0.5*3.5*0.488)

0.4

(2c/1) 0.9 0.5(=1.8–0.5*3.5*0.732)

0.9(=2.2–0.5*3.5*0.732)

-0.4

Trajectory based on evidence for 2008–12 (more significant spillovers)

Scenario 2012 2013 2014 Cumulative deviation of output from 2012 real GDP growth

level (2013–14), pptsBaseline (1) 0.9 1.8 2.2 2.2(2a/2) 0.9 0.9

(=1.8–0.5*3.5*0.498)1.3

(=2.2–0.5*3.5*0.498)0.4

(2b/2) 0.9 0.1(=1.8–0.5*3.5*0.996)

0.5(=2.2–0.5*3.5*0.996)

-1.2

(2c/2) 0.9 -0.8(=1.8–0.5*3.5*1.494)

-0.4(=2.2–0.5*3.5*1.494)

-3




Solvency: Illustrative Example for a Hypothetical Austrian BankStep 2: Simulation of the impact at the bank level (example for stylized bank)5

Change of key solvency parameters6

Scenario Loan impairment rates (Percent of credit exposure)

Preimpairment income (Percent of total capital)

2012 2013 2014 2012 2013 2014Baseline 0.5 0.4 0.4 10 10.3 10.52a/1 0.5 0.45 0.4 10 10.15 10.32b/1 0.5 0.5 0.45 10 10 10.12c/1 0.5 0.55 0.5 10 9.8 102a/2 0.5 0.5 0.45 10 10 10.12b/2 0.5 0.7 0.6 10 9.7 9.82c/2 0.5 0.9 0.8 10 9.2 9.5

Evolution of risk-weighted assets and capital7

Scenario RWAs (Indexed) Capital

2012 2013 2014 2012 2013 2014Baseline 100 90 90 10 10.58 11.212a/1 100 95 90 10 10.57 11.182b/1 100 100 95 10 10.56 11.152c/1 100 105 100 10 10.54 11.122a/2 100 100 95 10 10.56 11.152b/2 100 120 111 10 10.52 11.082c/2 100 140 132 10 10.47 10.99

Evolution of the bank’s capital ratioScenario Capital Ratio (= capital/RWA, percent)

2012 2013 2014Baseline 10.0 11.8 12.52a/1 10.0 11.1 12.52b/1 10.0 10.6 11.72c/1 10.0 10.0 11.12a/2 10.0 10.6 11.72b/2 10.0 8.8 9.92c/2 10.0 7.5 8.3

Source: Authors.Note: bps = basis points; GIIPS = Greece, Ireland, Italy, Portugal, and Spain; RWAs = risk-weighted assets; spec = specification.1The average impact of stress (in terms of GIIPS spreads) on euro area countries is 24 basis points (based on the panel analy-sis; see Appendix Table 12.1.1) and for Austria the relative severity of this impact approximately matches the impact ob-served for the EU, that is, it is 1.0 times of this level (GARCH analysis, average impact from 2008–12 based on Figure 12.2, panel 2, relative to average of the average impact for other EU countries).2The study used the higher impact on the GIIPS spreads from Appendix Table 12.1.1 and Appendix Table 12.1.2—that is, specification 2 (Appendix Table 12.1.1) and 1 (Appendix Table 12.1.2), respectively, that is, 24 bps and 49 bps, respectively.3The average impact of stress on euro area countries is 49 basis points (based on the panel analysis, Appendix Table 12.1.2), and for Austria the impact is again estimated to be at a similar level (GARCH analysis, average impact from 2008–12, Figure 12.2, panel 2, relative to average of the average impact for other EU countries).4The GDP elasticities of sovereign debt spreads vary between 0.5 (for example, Brazil) and 3.5.5See Hardy and Schmieder 2013 for further information.6Credit growth is assumed to be constant for simplification.7For simplification, RWA elasticity to credit losses is assumed to be 0.5—that is, for a 1-percentage-point change of credit loss rates RWAs will change by 0.5 percentage points.


Appendix 12.5.Illustrative Example for Liquidity

This appendix provides an illustrative example for a hypothetical bank in Austria.Step 1. GDP trajectory for Austria, adjusted for the impact of spillovers.The first steps uses the same GDP trajectories as for solvency (see Appendix 12.4). Accordingly, the severity of the liquidity

shock is simulated relative to the Lehman Brothers benchmark scenario in Appendix 12.4. Specifically, based on the observa-tion that the cumulative US real GDP growth deviated by about 8 percentage points from the long- term average, the corre-sponding figures are computed for each of the scenarios. For Austria (and for the other European countries), the baseline growth rates for 2013–14 (that is, 2 percent) are (for simplicity) used as a proxy for the long- term trend. For Scenario 2c/2, the cumulative deviation from the baseline is 5.2 percentage points. For the severity of the liquidity test, the study therefore used the stress parameters for the severe scenario in Appendix 12.3 multiplied by a factor of 0.65 (=5.2/8).

Step 2. Simulation of the impact at the bank level (example for stylized bank).27

Relevant asset and liability balance sheet items are shocked based on the severity of each scenario, that is, the stress factor (such as 0.65) multiplied by the respective stress parameters. The balance sheet items are taken from BankFocus. For the long- term refinancing operation, the available total funding was assigned to the single banks based on their size, using the available evidence for the total at the country level.

In the table below, Scenario 2c/2 is simulated for a stylized bank based on Austria. The composition of the banks’ asset and liabilities resemble those of an average Organisation for Economic Co- operation and Development (OECD) bank.28 The stress factor reduces the haircuts and outflows of the benchmark scenario. In the example, the bank is able to generate an inflow of 21.5 units of assets, compared to a required level of 13.7 units, whereby the bank remains liquid.

27 See Schmieder and others 2012 for further information.28 See Schmieder and others 2012, p. 38, for more information.


Liquidity: Illustrative Example for a Hypothetical Austrian BankAssets (of stylized bank)

Portion of Total1 Haircut, Percent (Appendix 12.3)

Haircut Scenario 2c/2

Available Assets (Fire sales)

Cash and cash-like 4 0 0 4.0Government securities 6 5 3 5.8Trading securities 5 30 20 4.0Other securities 15 75 49 7.7Loans 60 NA NAOther 10 NA NA

Liabilities (of stylized bank)

Portion of Total2 Outflow, Percent (Appendix 12.3)

Outflow Scenario 2c/2

Required Funding

Customer term deposits 30 10 6.5 2Customer demand deposits 20 20 13 2.6Secured short-term wholesale

funding10 20 13 1.3

Unsecured short-term wholesale funding

10 100 65 6.5

Long-term funding 20 0 0 0Equity-based funding 10 0 0 0Contingent liabilities 20 10 6.5 1.3

Source: Authors.1Aligned to the average composition of Organisation for Economic Co-operation and Development banks‘ balance sheets. See Schmieder and others 2012, p. 38.2Aligned to the average composition of Organisation for Economic Co-operation and Development banks‘ balance sheets. See Schmieder and others 2012, p. 38.



Frank, Nathaniel, and Heiko Hesse. 2009. “Financial Spillovers to Emerging Markets during the Global Financial Crisis.” IMF Working Paper 09/104, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2016 /12/31/ Financial- Spillovers- to- Emerging- Markets- During -the-Global-Financial-Crisis-22936.

Hardy, Daniel C., and Christian Schmieder. 2013. “Rules of Thumb for Bank Solvency Stress Testing.” IMF Working Paper 13/232, International Monetary Fund, Washington, DC. https://www .imf.org/en/Publications/WP/Issues/2016/12/31/Rules-of -Thumb-for-Bank-Solvency-Stress-Testing-41047.

Hesse, Heiko, Ferhan Salman, and Christian Schmieder. 2014. “How to Capture Macro- Financial Spillover Effects in Stress Tests?” IMF Working Paper 14/103, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2016 /12/31/ How- to- Capture- Macro- Financial- Spillover-Effects -in-Stress-Tests-41644.

International Monetary Fund (IMF). 2011. “Euro Area Policies: Spillover Report for the 2011 Article IV Consultation and Se-lected Issues.” IMF Country Report 11/185, Washington, DC. http://www.imf.org/external/pubs/cat/longres.aspx?sk=25056.0.

———. 2012a. Global Financial Stability Report— The Quest for Lasting Stability, Chapter 3. Washington, DC, April. https://www.imf.org/external/pubs/ft/gfsr/2012/01/.

———. 2012b. “2012 Spillover Report.” IMF Policy Paper, Wash-ington, DC. https://www.imf.org/en/Publications/Policy-Papers /Issues/2016/12/31/2012-Spillover-Report-PP4678.

———. 2013. “European Union: Stress Testing of Banks— Technical Note.” IMF Country Report 13/68, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / European- Union- Publication- of- Financial- Sector-Assessment -Program-Documentation-Technical-40396.

Jobst, Andreas A., and Dale F. Gray. 2013. “Systemic Contingent Claims Analysis— Estimating Market- Implied Systemic Sol-vency Risk.” IMF Working Paper 13/54, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/ Systemic- Contingent- Claims- Analysis - Estimating-Market-Implied-Systemic-Risk-40356.

Jobst, Andreas, Li Lian Ong, and Christian Schmieder. 2013. “A Framework for Macroprudential Bank Solvency Stress Testing: Application to S- 25 and Other G20 Country FSAPs.” IMF Working Paper 13/68, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/ A- Framework- for- Macroprudentia l- Bank- Solvency- Stress- Testing- Application- to- S-25-and-Other-G -40390.

Oura, Hiroko, and Liliana Schumacher. 2012. “ Macro- Financial Stress Testing— Principles and Practices.” IMF Policy Paper, Washington, DC, August. https://www.imf.org/en/Publications / Policy- Papers/Issues/2016/12/31/ Macrof inancial- Stress -Testing-Principles-and-Practices-PP4702.

Schmieder, Christian, Heiko Hesse, Benjamin Neudorfer, Claus Puhr, and Stefan W. Schmitz. 2012. “Next Generation System- Wide Liquidity Stress Testing.” IMF Working Paper 03/12, International Monetary Fund, Washington, DC. https://www .imf.org/external/pubs/cat/longres.aspx?sk=25509.0.

Schmieder, Christian, Claus Puhr, and Maher Hasan, 2011, “Next Generation Balance Sheet Stress Testing.” IMF Working Paper 11/83, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Next -Generation-Balance-Sheet-Stress-Testing-24798.

REFERENCESAikman, David, Piergiorgio Alessandri, Bruno Eklund, Prasanna Gai,

Sujit Kapadia, Elizabeth Martin, Nada Mora, Gabriel Sterne, and Matthew Willison. 2009. “Funding Liquidity Risk in a Quantita-tive Model of Systemic Stability.” Bank of England Working Paper 372, Bank of England, London, United Kingdom. https://www .bankofengland.co.uk/ working- paper/2009/ fundingliquidity - risk- in- a-quantitative-model-of-systemic-stability.

Austrian National Bank (OeNB). 2013. “ARNIE in Action: The 2013 FSAP Stress Tests for the Austrian Banking System.” Fi-nancial Stability Report 26, Vienna, Austria.

Barnhill, Theodore Jr., and Liliana Schumacher. 2011. “Modeling Correlated Systemic Liquidity and Solvency Risks in a Financial Environment with Incomplete Information.” IMF Working Pa-per 11/263, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 / Modeling- Correlated- Systemic- Liquidity- and- Solvency- Risks - in-a-Financial-Environment-with-25356.

Basel Committee on Banking Supervision (BCBS). 2008. Princi-ples for Sound Liquidity Risk Management and Supervision. Ba-sel: Bank for International Settlements. https://www.bis.org /publ/bcbs138.htm.

———. 2013. “Liquidity Stress Testing: A Survey of Theory, Em-pirics, and Current Industry and Supervisory Practices.” BCBS Working Papers 24, October, Bank for International Settle-ments, Basel, Switzerland. https://www.bis.org/publ/bcbs_wp24 .htm.

Bollershev, Tim. 1990. “Modelling the Coherence in Short- Run Nominal Exchange Rates: A Multivariate Generalized ARCH Approach.” Review of Economics and Statistics 72 (3): 498–505.

Borio, Claudio, Matthias Drehmann, and Kostas Tsatsaronis. 2012. “ Stress- Testing Macro Stress Testing: Does It Live up to Expectations?” BIS Working Paper 369, Bank for International Settlements, Basel, Switzerland. https://www.bis.org/publ /work369.htm.

Caceres, Carlos, and D. Filiz Unsal. 2011. “Sovereign Spreads and Contagion Risks in Asia.” IMF Working Paper 11/134, Interna-tional Monetary Fund, Washington, DC. https://www.imf.org /en/Publications/WP/Issues/2016/12/31/ Sovereign- Spreads - and-Contagion-Risks-in-Asia-24910.

Čihák, Martin. 2007. “Introduction to Applied Stress Testing.” IMF Working Paper 07/59, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ Introduction-to-Applied-Stress-Testing -20222.

Engle, Robert. 2002. “Dynamic Conditional Correlation: A Sim-ple Class of Multivariate Generalized Autoregressive Condi-tional Heteroskedasticity Models.” Journal of Business & Economic Statistics 20 (3): 339–50.

Foglia, Antonella. 2009. “Stress Testing Credit Risk: A Survey of Authorities’ Approaches.” International Journal of Central Banking (September). http://www.ijcb.org/journal/ijcb09q3a1 .htm.

Frank, Nathaniel, Brenda González- Hermosillo, and Heiko Hesse. 2008. “Transmission of Liquidity Shocks: Evidence from the 2007 Subprime Crisis.” IMF Working Paper 08/200, Inter-national Monetary Fund, Washington, DC. https://www.imf .org/en/Publications/WP/Issues/2016/12/31/ Transmission- of - Liquidity- Shocks- Evidence- from-the-2007-Subprime-Crisis -22238.




https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Rules-of-Thumb-for-Bank-Solvency-Stress-Testing-41047





http://www.imf.org/external/pubs/cat/longres.aspx?sk

https://www.imf.org/external/pubs/ft/gfsr/2012/01/.


https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/2012-Spillover-Report-PP4678.

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/2012-Spillover-Report-PP4678.




















https://www.bis.org/publ/bcbs_wp24.htm

https://www.bis.org/publ/bcbs_wp24.htm







http://www.ijcb.org/journal/ijcb09q3a1.htm













Vitek, Francis, and Tamim Bayoumi. 2011. “Spillovers from the Euro Area Sovereign Debt Crisis: A Macroeconometric Model Based Analysis.” CEPR Discussion Paper 8497, Center for Eco-nomic and Policy Research, Washington, DC. https://cepr.org /active/publications/discussion_papers/dp.php?dpno=8497.

Wong, Eric, and Cho- Hoi Hui. 2009. “A Liquidity Risk Stress Testing Framework with Interaction between Market and Credit Risks.” HKMA Working Paper 06/2009, Hong Kong Monetary Authority, Hong Kong SAR. https://www.hkma .gov.hk/eng/publications-and-research/research/working -papers/2009/.

Taleb, Nassim N., Elie Canetti, Tidiane Kinda, Elena Lou-koianova, and Christian Schmieder. 2012. “A New Heuristic Measure of Fragility and Tail Risks: Application to Stress Test-ing.” IMF Working Paper 216/2012, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/ A- New- Heuristic- Measure- of- Fragility - and- Tail- Risks-Application-to-Stress-Testing-26222.

Van den End, Jan Willem. 2008. “Liquidity Stress Tester: A Macro Model for Stress- Testing Banks’ Liquidity Risk.” DNB Working Paper 175/2008, Dutch National Bank, Amsterdam, Netherlands. https://www.dnb.nl/en/news/ dnb- publications/ dnb- working -papers-series/dnb-working-papers/auto175528.jsp.



https://www.hkma.gov.hk/eng/publications-and-research/research/working-papers/2009/




https://www.imf.org/en/Publications/WP/Issues/2016/12/31/A-New-Heuristic-Measure-of-Fragility-and-Tail-Risks-Application-to-Stress-Testing-26222



https://www.dnb.nl/en/news/\dnb-\publications/\dnb-\working-papers-series/dnb-working-papers/auto175528.jsp

https://www.dnb.nl/en/news/\dnb-\publications/\dnb-\working-papers-series/dnb-working-papers/auto175528.jsp

CHAPTER 13

Real and Financial Vulnerabilities from Cross- Border Banking Linkages

KYUNGHUN KIM • SROBONA MITRA

This chapter looks at the vulnerabilities stemming from banking sector linkages between countries and their macroeconomic effects. It finds that credit risks (from a banking system’s claims on other countries) and funding risks (from a banking system’s liabilities to another) declined be-

tween 2008 and 2012. It also finds that funding vulnerabilities have real effects. During normal times, funding vulnerabilities are associated with significant positive GDP growth surprises. During crisis times, funding vulnerabilities are associated with significant negative GDP growth surprises. The results imply that policymakers should pay more attention to understanding cross- border funding risks.

The chapter explores the financial risks of cross- border banking linkages using network analysis. Rather than just identifying and quantifying linkages, the chapter simulates the impact on capital levels of the credit and funding shocks that could be transmitted through direct and indirect (dom-ino effect) banking linkages. The chapter examines whether the potential impact on capital— summarized by vulnerabil-ity indices— has changed in the last five years. Using net-work analysis ( Espinosa- Vega and Solé 2011), the chapter shows the trends in the financial systems’ vulnerability to network effects of shocks on either side of the balance sheet.

The chapter then asks whether the vulnerability of a bank-ing system from interconnections influences output. For the network analysis to have macro- financial implications, the real effects of higher vulnerability to network shocks are esti-mated using an econometric model. Specifically, a set of panel fixed effect regressions examine the relationship be-tween vulnerability to cross- border credit or funding shocks and GDP growth rate surprises, measured by the difference between actual GDP growth and Consensus Forecasts.

1. INTRODUCTIONThe global financial crisis made it clear that financial shocks could be quickly transmitted through global banks. The tightly interconnected financial systems were put through several tests during the crisis. The banking linkages, by far the largest and the deepest segment of financial flows, saw reduced flows.

Against this backdrop, this chapter asks two questions. First, how have countries’ vulnerabilities arising from bank-ing network linkages changed? The chapter examines two kinds of risks— credit risk and funding risk. These risks are related to the nature of the interlinkages— credit risks mate-rialize through a banking system’s claims on other countries, and funding risks arise through banking system’s liabilities to one another. The vulnerabilities are related to both expo-sures to these risks and the capital buffers available against these risks. Second, what are the macroeconomic effects of these vulnerabilities? That is, are these specific vulnerabili-ties associated with real GDP growth beyond what is ex-pected in macroeconomic forecasts?

This chapter is based on IMF Working Paper 14/136 (Kim and Mitra 2014). The authors would like to thank, without implicating, James Morsink, who suggested the project and provided extensive and useful comments; Eugenio Cerutti, Jiaqian Chen, Allison Holland, Gian Maria Milesi- Ferretti, Prachi Mishra, Lam Nguyen, Ratna Sahay, Jason Weiss, and Hongyan Zhao for other helpful suggestions; and Juan Sole and Marco Espinosa- Vega for making available the codes for network analysis through an Excel add- in; and various participants in internal seminars at the IMF.


Real and Financial Vulnerabilities from Cross-Border Banking Linkages300

lance. By simulating credit or funding shocks, they obtain vulnerability indices for each banking system. Using this tool, Espinosa- Vega, Solé, and Kahn (IMF 2010) also pro-pose a framework for capital requirements for those banks that have a large contribution to systemic risk in a network. Cerutti, Claessens, and McGuire (2011) highlight data needed for properly analyzing contagion risk, an exercise similar in spirit to the network analysis, and Cerutti 2013 proposes two new measures for better capturing creditor banking systems’ foreign credit exposures and borrower countries’ reliance on foreign bank credit, by combining Bank for International Settlements (BIS) data with bank- level data.

This is the first work to distinguish between cross- border risks arising from the asset and the liability side of the bank-ing system’s balance sheet and relate these different risks to macroeconomic effects. The chapter applies the methodol-ogy proposed by Espinosa- Vega and Solé (2011) to a dataset that covers 20 countries over 2006–12 and shows the real impacts for the countries receiving the shocks. The chapter documents how the cross- border vulnerabilities of the bank-ing system have evolved since 2006 and shows how the vul-nerability index from the network analysis is associated with output shocks during normal times and crises.

3. DATA AND METHODOLOGYThe vulnerability from interconnections goes beyond the simple mapping of exposures between countries. The vulner-ability or susceptibility to network effects is measured by the potential capital shortfall in the event of a tail risk in which one banking system fails. It is measured by the average change in the capital level in percentage of the preshock capital due to the direct and domino effects of every banking system failing. Therefore, the vulnerability of any country to a shock in another banking system depends upon four fac-tors: effects through direct bilateral links, domino effects through indirect network links, own capital levels, and capi-tal levels in the major shock- propagating countries. Vulner-ability goes up with stronger banking bilateral links and gets magnified by domino effects running through link-of-links. Lower capital buffers in the shock recipients, as well as in shock propagators, increases vulnerability in any given coun-try. Of course, the use of aggregate data might not capture potential systemic vulnerabilities arising from individual large institutions.

Data

To run the network analysis, data is needed on the matrix of exposures between countries. This means a banking system’s credit (claims) and liabilities vis- à- vis another country’s banking system are needed. The BIS consolidated banking statistics is used for the purpose. Since it does not have data on cross- border liabilities of banking systems, this is proxied by looking at the claims of the counterparty banking

The chapter has two main findings. First, vulnerabilities of banking systems to both credit and funding risks have declined since the crisis. This decline is due to both lower exposures and increases in capital for the global banking sys-tem. Second, funding vulnerabilities have real effects. Dur-ing normal times, funding vulnerabilities are positively associated with GDP growth surprises; during crises, the same vulnerabilities exacerbate the negative GDP growth surprises. Credit vulnerabilities, on the other hand, are not associated with GDP surprises.

The rest of the chapter is organized as follows. The related literature is discussed in Section 2; the methodology and the data in Section 3; the findings on the vulnerability trends and the association of the vulnerabilities with GDP growth surprises are discussed in Sections 4 and 5, respectively; and Section 6 concludes.

2. RELATED LITERATUREThe chapter builds on the recent literature on cross- border financial interconnectedness and its implications for financial stability and real output. Kalemli- Ozcan, Papaioannou, and Perri 2013 find that higher banking linkages are associated with more divergent output cycles during normal times; however, this relationship becomes weaker during financial crisis. Abiad and others 2013 distinguish between tradi-tional financial linkages and common shocks to show that output comovement across countries— synchronized output collapses— occurs during financial crises through common shocks. Cetorelli and Goldberg 2010 show how the US fi-nancial crisis was transmitted to other countries through the relationship between multinational banks and their foreign affiliates. Albertazzi and Bottero 2014 suggest that the for-eign banks restricted credit supply more than their domestic counterparts, using disaggregated Italian bank- firm data. De Haas and Van Lelyveld (2014), Giannetti and Laeven (2012), and Popov and Udell (2012) have empirical evidence to show that multinational banks restricted credit supply in the host countries during the financial crisis.

Cihák, Munoz, and Scuzzarella 2011 show an M- shaped relationship between the financial stability of a country’s banking sector and its cross- border interconnectedness mea-sured by network centrality measures— starting from low integration, increases in global interconnectedness for the banking system are associated with a reduced probability of a banking crisis. For a banking system whose interconnect-edness is over a certain value, increases in interconnectedness can increase the probability of a banking crisis. Relatedly, Minoiu and others 2015 show that increases in a country’s own connectedness and decreases in its neighbors’ connect-edness are associated with a higher probability of banking crises. Nier and others 2007 investigate how systemic risk is affected by the structure of the banking system, using net-work models.

Espinosa- Vega and Solé 2011 show that network analy-sis can be used as a tool for cross- border financial surveil-


Kyunghun Kim and Srobona Mitra 301

said to fail, and these then trigger domino impacts on all oth-ers. The simulation goes on until there are no more failures.

For funding risk (panel 2), if B fails, it is unable to roll over ρ (the assumption is 0.5 in the baseline) times other countries’ liabilities, including A’s. A, and other countries, then try to fire sell their assets at a haircut (the assumption is half, which translates into δ = 1) and take a hit on capital. If it fails, this triggers further failures. Again, the domino goes on until there are no more failures.

The network model produces vulnerability indices. The index is simply the average capital depletion if other banking systems fail. This number is derived by running the network model for each country, at each point of time, 2005–12, separately for credit risk and funding risk. So, there is a credit vulnerability index and a funding vulnerability index for each country. Then there is a global index (for all 20 countries) that takes a weighted average of the indices for each country, weighted by the sum of gross credit and liabili-ties of each country.

The vulnerability index has a practical meaning. The credit index indicates the potential capital loss (in percent of preshock capital) of a banking system’s opening up to for-eign expansions, increasing foreign claims, or not having ad-equate capital buffers against those claims. The funding index gives information on the potential capital loss rate of a banking system due to opening up to higher foreign funding (liabilities risk) without adequate capital buffers to with-stand fire sales if necessary. The index itself is influenced by four factors for given levels of the parameters, λ, ρ, and δ : direct linkages, indirect linkages, own capital levels, and those of others.

Panel fixed effects regressions are used to look at the association between GDP growth surprises and the vul-nerability indices, for 20 countries, for seven years 2006–12. The GDP growth surprises are calculated by taking the difference between actual GDP growth and the fore-cast of GDP growth made in the previous December by consensus economics. The average growth surprises for the 20 countries show the large negative surprises during the crisis years 2008 and 2009 (Figure 13.2). The regressions take the growth surprise as the dependent variable, and regresses it on a dummy variable that takes the value of 1 for the two crisis years, the vulnerability index, and a term that interacts the vulnerability index with the crisis dummy (see equation 13.1).

= α +β + γ+λ + ε

−

=

− −

− −

−

y crisis VULVUL crisis

y

crisis

it i itv

itv

it

v

ˆ * ** *

where, ˆ is GDP Growth Surprise =Actual Real GDP Growth Consensus Forecast

VUL : Vulnerability index for Credit risk or Fundingrisk 1 for 2008 and 2009

2008 09 1

1 2008 09

2008 09 (13.1)

If the cross- border credit and funding risks are well under-stood by macroeconomic forecasters, the indices would not be expected to affect the growth surprises. This is because the

systems. The liabilities side, therefore, is measuring the lia-bilities of all sectors of the economy to BIS reporting banks with headquarters in another country. Even though it is im-precise, it is assumed that most of these liabilities are sourced through the banks and measure the banking system’s indi-rect liabilities to the BIS reporting banks in the other coun-try. This is the best that can be done with the published data, which is available for 20 countries.1

There is a 20 by 20 matrix for each of the years 2005 through 2012. For instance, in 2008, the US banks lent $268 billion to the United Kingdom and the United States (all sectors) borrowed $1.217 trillion from the United King-dom. By 2012, the United States lent more than twice to the United Kingdom and borrowed less from the Kingdom (Table 13.1).

In order to understand the vulnerabilities from cross- border exposures, one needs to weigh the exposures against the financial buffers. So, data on capital was needed, which was obtained from Bankscope. The sum was taken of the capital that each banking system’s commercial banks, sav-ings banks, cooperative banks, real estate and mortgage banks, investment banks, other nonbanking credit institu-tions, and specialized governmental credit institutions own. A wide net was cast to capture data on capital from as many institutions residing and headquartered in a country as was possible to get a sense of buffers.

Methodologies

Deriving vulnerability indices based on network analysis

The network model used in this chapter was developed in Chapter 2 of the IMF’s April 2009 Global Financial Stability Report (IMF 2009) and described in Espinosa- Vega and Solé 2011. The model runs simulations using the data on expo-sures and capital. Specifically, it lets each banking system fail and calculates the impact of the credit risk from such a failure on other banking systems’ capital. Similarly for fund-ing risk. There are both direct and domino effects of a banking system’s failure on others.

The method can be illustrated by means of a stylized bal-ance sheet of a banking system, say A (Figure 13.1). For credit risk (panel 1), if, another banking system B’s banks fail due to some unexplained event, it is unable to repay λ (the assump-tion is 0.5 in the baseline) of its dues to all other countries. These assets then go “bad” for all the creditor banking sys-tems, A is one of them, and these should have sufficient capital to absorb this loss. If they don’t, then the banking systems are

1 Confidential bilateral data based on the BIS Locational Statistics, which was available for the third quarter of 2013, provides the break-down by bank and nonbank exposures. On average, 60 percent of the cross- border claims of the BIS reporting banks resident or located in a certain country are on the banking sector; the average is higher for the G7 countries.


Real and Financial Vulnerabilities from C

ross-Border Banking Linkages302

TABLE 13.1

Capital and Financial Exposure between Banking Systems (In millions of US dollars, column countries’ claims on rows)2008:Q4

Capital Country AU AT BE CA FI FR DE GR IN IE IT JP NL PT ES SE CH TR GB US

41,858 Australia 2124 5065 10350 NaN 43577 50961 35 522 7727 1409 52082 72562 593 3597 2572 21597 24 100324 45287

56,095 Austria 1191 5117 1358 NaN 20522 105679 173 231 5806 146042 6757 9203 840 3100 1815 13724 425 12487 4550

39,476 Belgium 2779 2985 5055 NaN 111135 41295 173 997 8203 11687 21091 152364 1828 12203 4122 18067 778 43500 26415

59,702 Canada 6439 1849 7263 NaN 26155 41701 141 2307 12385 2496 45193 39731 284 1656 2530 19071 19 80756 66152

3,615 Finland 503 1068 2746 218 6676 14044 7 26 NaN 1665 6357 4021 47 1755 107380 2945 2 6409 3880

368,163 France 10738 10879 125644 12430 NaN 193246 673 863 25496 55253 124562 124615 7261 46326 11399 66233 1260 241107 77068

183,041 Germany 11372 52427 58893 12242 NaN 279538 2228 1972 49714 337447 158334 174138 9854 47118 74498 113691 5190 159978 92620

15,083 Greece 480 5617 10175 395 NaN 75224 38389 19 8480 9513 6176 12868 6376 1012 1310 69552 135 12713 6753

52,769 India 1473 761 5397 NaN NaN 11355 19498 40 NaN 764 13168 22020 157 1155 307 5319 0 49672 38313

30,583 Ireland 1651 5097 45550 14233 NaN 68115 202202 323 263 24439 23857 35438 3781 14832 5247 20387 98 190440 33014

154,393 Italy 10490 17628 51951 2334 NaN 468850 207194 278 486 46537 48270 66955 3483 48680 3912 19946 763 74839 25526

363,573 Japan 2397 486 2797 6225 NaN 218920 65619 12 716 17228 NaN 28170 21 1214 943 121091 211 113158 123333

108,956 Netherlands 6998 12068 85453 10545 NaN 128186 167279 807 964 17550 24364 45822 8985 21375 9352 51784 2840 129601 52599

18,026 Portugal 318 2539 12040 NaN NaN 29918 44492 40 49 6341 6197 3056 13842 77424 569 7524 3 21952 1848

141,955 Spain 2107 7919 43964 3050 NaN 176421 253676 265 166 33704 28463 25711 124773 28655 7106 20360 245 124572 33458

54,611 Sweden 725 1611 2581 1349 NaN 16154 37935 17 170 6173 1988 14505 8197 901 1917 7810 80 16425 8323

105,385 Switzerland 3551 10923 9840 3877 NaN 57483 67469 454 307 7415 11342 24846 18879 2250 5436 5545 477 44909 22865

18,582 Turkey 104 2517 15610 NaN NaN 12355 16072 18317 105 NaN NaN 3386 21229 906 131 300 4159 17340 12806

468,068 United Kingdom

103958 23560 127774 64156 NaN 394557 509133 6965 3600 222201 49898 164072 180607 7639 349916 38713 219202 3418 268187

1,088,470 United States

41931 21060 113161 430465 NaN 766345 640501 3953 7789 122477 32711 911642 335920 8882 132623 42703 827133 5136 1217127

(continued)


Kyunghun Kim and Srobona M

itra303


Capital and Financial Exposure between Banking Systems (In millions of US dollars, column countries’ claims on rows)

2012:Q4

Capital Country AU AT BE CA FI FR DE GR IN IE IT JP NL PT ES SE CH TR GB US

43,058 Australia 792 1957 22219 369 18053 23724 69 1043 715 2265 130606 79332 24 2515 2352 28206 28 72924 115419

71,233 Austria 279 730 1096 320 14618 75685 963 57 315 101145 6170 9839 124 4824 1933 8570 198 7819 11998

48,663 Belgium 882 1550 2331 581 222983 27924 276 1153 524 4427 19291 116618 441 5226 2378 6374 163 18016 18469

103,774 Canada 18838 982 1340 150 16901 27668 166 2799 371 3833 62639 10056 168 1695 2467 21860 11 104838 129360

5,107 Finland 864 869 452 1948 7434 18404 168 34 NaN 1200 4400 5687 55 2014 151832 4515 1 10263 12242

462,704 France 8251 11727 25319 27236 2715 195139 1670 754 5227 45764 166368 67377 6990 31531 8030 57918 957 221012 213807

237,477 Germany 21718 43746 12749 25595 2706 197643 3510 2078 2299 237073 145811 185007 2159 57395 80606 71785 3057 273571 217456

8,310 Greece 95 331 32 NaN NaN 2798 5293 4 113 902 404 2343 7400 779 76 1527 90 5631 3201

93,048 India 9606 382 404 5852 NaN 15422 23581 1 NaN 2297 25422 13876 13 290 256 10938 6 84264 80077

31,989 Ireland 2531 1399 20000 4591 399 37954 81581 405 101 10207 23486 12726 3978 6047 1560 14735 2 121975 46515

174,556 Italy 578 15675 10339 4636 274 343207 129200 520 146 955 37068 33068 2839 27740 1433 19981 62 49227 42716

275,686 Japan 26875 NaN 795 15745 2 92689 45681 104 578 184 NaN 8029 22 3092 956 68161 913 130355 372517

153,016 Netherlands 9960 8127 23600 13282 2295 158134 157528 3095 1381 2235 19810 65250 9392 19351 10501 35214 2636 171369 106970

22,274 Portugal 132 791 716 NaN 139 16916 21670 17 40 473 1725 1178 4500 71567 190 1503 0 17337 4765

189,093 Spain 891 3018 9471 2513 433 108033 120717 218 103 4085 22977 21221 53686 22674 2792 17550 208 82863 49436

61,728 Sweden 2525 1539 605 2727 3462 21303 34873 88 157 530 2167 20939 7557 159 2594 9583 13 15470 27889

159,268 Switzerland 8220 9696 1323 5627 701 68058 60645 854 655 849 10790 29162 13645 2065 7347 3185 271 86465 76823

36,178 Turkey 380 1811 1321 2517 NaN 30847 19018 31083 107 NaN 6313 8390 20782 2 21455 215 6147 37528 24492

761,648 United Kingdom

133427 17012 25875 107694 2312 224666 409259 11866 5382 111414 49770 188909 128906 5246 406941 48872 166389 3090 634309

1,622,337 United States

107814 10760 20742 720340 461 402553 496792 4236 9389 7107 30977 1296100 164675 5038 203576 103150 670629 4819 1080697

Sources: Bank for International Settlements; Bankscope; and authors’ calculations.Note: AT = Austria; AU = Australia; BE = Belgium; CA = Canada; CH = Switzerland; DE = Germany; ES = Spain; FI = Finland; FR = France; GB = United Kingdom; GR = Greece; IE = Ireland; IN = India; IT = Italy; JP = Japan; NL = Netherlands; PT = Portugal; SE = Sweden; TR = Turkey; US = United States.



4. IS THE WORLD SAFER FROM CROSS- BORDER BANKING LINKAGES?The matrix of banking exposures across countries reveals notable changes between 2008 and 2012 (Table 13.1). The financial exposures and funding of non- European countries are on the rise, especially of Canada, Japan, and the United States. The euro area countries have all seen a drop in both cross- border exposures and funding; this is especially so for France and Germany. This phenomenon, often called “fragmentation,” has left policymakers worried about the cost of funds and the availability of credit in European countries. Whether the world is safer from cross- border banking connections depends upon bilateral exposures, network exposures through dom-ino effects, and on own and other countries’ capital levels.

Vulnerability of the overall global banking system to net-work shocks was high before 2008 (Figures 13.3 and 13.4). Going back to 2006, about 25–30 percent of capital, on an av-erage in a country could have been impaired due to network

GDP forecasts would already take into account the risks that could affect a country through the cross- border banking channels so that the residuals, the GDP growth surprises, should not be correlated with information available at the time of making these forecasts.

To check if data on overall exposures (foreign claims + foreign liabilities) and capital, separately would have de-livered similar results, obviating the need to run the net-work analysis, a second set of regressions using these components was added, instead of the vulnerability indi-ces (equation 13.2). If higher exposures and lower capital helped explain growth surprises, then understanding these components of the network analysis would be beneficial by themselves.

(13.2)

= α +β + γ + λ+ γ + λ + ε

− − − −

− − −

y crisis Cap Cap crisisExp Exp crisis

it i it it

it it it

ˆ * * * ** * *

where Cap: Capital/GDPExp: (Foreign claims + Foreign liabilities)/GDP

2008 09 1 1 1 1 2008 09

2 1 2 1 2008 09

ai

�xhi �xhi

di

ki

Assets Liabilities

�xjij

�xijj

ai

(1+�)�xih��xih�xih

di

ki

Assets Liabilities

�xjij

�xijj

1. Credit Shock 2. Funding Shock

Source: April 2009 Global Financial Stability Report, Chapter 2 (IMF 2009); and Sole and Espinosa-Vega 2010.Note: A “ ” represents the amount by which capital, k, will be hit in the first round. x = cross-border credit and funding; a = other assets; d = other liabilities, like customer deposits and debt; k = capital; λ = fraction of interbank loans that does not get repaid (0.50 in the baseline); ρ = fraction of interbank liabilities that does not get rolled over (0.50 in the baseline); δ = haircut on interbank assets that need to be fire sold to replace the fraction of interbank funding that is not rolled over (1 in the baseline).

Figure 13.1 Credit Shock and Funding Shock Illustrated with Stylized Banking System Balance Sheets

–3

1

–2

–1

0

072006 08 09 1210 11

Source: Consensus Forecasts.Note: GDP growth rate surprise = actual GDP growth rate (WEO) – GDP growth rate forecast (Consensus Forecasts, average of the GDP growth rate forecasted over the previous December).

Figure 13.2 Growth Rate Surprise(Average difference between actual and forecasted GDP for 20 countries, in percentage points)



constant at the 2008 levels (Figures 13.3 and 13.4). Even after adjusting for capital, the vulnerability indices (weighted by total exposures of countries) have trended down for both credit and funding shocks, which suggests that the actual strength and number of interconnections had also fallen.

The aggregate results mask wide cross- country differences in vulnerability trends on credit shocks. There are three groups of countries depending upon whether vulnerabilities on cross- border assets have trended down or up or largely remained unchanged between 2008 and 2012 (Figure 13.5):

• Belgium and Ireland started from high levels of suscep-tibility to shocks on their cross- border investments, and these have come down significantly. The downward trend was mainly attributable to a lower volume of cross- border investments than to higher capital levels. In addition, the United Kingdom, France, Italy, Ger-many, Switzerland, and other coun tries (shown in the middle of Figure 13.5) also experienced downward trends.

effects of credit and funding shocks. Since then, countries’ sus-ceptibility to these shocks started coming down until 2008, and then fell after that. The decrease until 2008 was mostly due to the lower volume of flows between advanced countries since mid- 2007. The vulnerabilities in 2006, based on pub-lished balance sheet data on the banking network, could have served as an early warning on the extent of losses that banking systems would suffer if there were to be an extreme event.

Since the end of 2008, banking systems have been generally less vulnerable to ripple effects from network shocks due to two reasons. With the collapse of Lehman Brothers and the sever-ance in some linkages due to the materialization of the adverse shocks, individual banking systems now had lower volume of inflows through banks. And capital levels had increased on the aggregate after the crisis so that for any inflow the buffers were greater across countries, in general, to absorb the shocks.

To show that higher buffers were not entirely responsible for the lower vulnerability levels, the network analysis is re-peated for 2009–12, assuming that the capital levels are

Actual capitalFixed capital at 2008:Q4

0

15

30

20

25

062004 08 10 12

Inde

x of

vul

nera

bilit

y

Time

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.1The index of vulnerability shows the percentage of capital impairment in a banking system due to the failure of other banking systems. The aggregate index shown above is the weighted average of the vulnerability indices of the 20 countries in the sample, weighted by the country’s total financial exposure.

Figure 13.3 Vulnerability to Credit Shock1

(Financial exposure weighted average of vulnerability; global banking system, 2005–12)

Actual capitalFixed capital at 2008:Q4

10

15

30

20

25

062004 08 10 12

Inde

x of

vul

nera

bilit

y

Time

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.1The index of vulnerability shows the percentage of capital impairment in a banking system due to the failure of other banking systems. The aggregate index shown above is the weighted average of the vulnerability indices of the 20 countries in the sample, weighted by the country’s total financial exposure.

Figure 13.4 Vulnerability to a Funding Shock1

(Financial exposure weighted average of vulnerability; global banking system, 2005–2012)



Interestingly, higher capital buffers seem to have largely contributed toward lower vulnerability to funding shocks, especially for two emerging economies for which there is published data. For India and Turkey and some larger coun-tries, vulnerability to funding shocks came down since the

• In Greece, the susceptibility to network credit effects of cross- border investments increased over time.

• The United States, Japan, Canada, Australia, India, and Turkey are some countries in the middle, where cross- border credit risks did not change significantly.

Actual capital Fixed capital at 2008:Q4

Finland

2004 1206 08 10 2004 1206 08 10 2004 1206 08 10 2004 1206 08 10 2004 1206 08 10

Time

0

40

102030

Australia

0

40

102030

France

0

40

102030

Germany

0

40

102030

Greece

Spain

Netherlands

Canada

United States

United Kingdom

Switzerland

Turkey

India

Portugal

Belgium

Sweden

Japan

Austria

Ireland

Italy

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.Note: Foreign claims of Finland are available after 2010. The graphs are placed in order of difference between 2008:Q4 and 2012:Q4 (ascending).

Figure 13.5 Individual Banking System’s Vulnerability to the Credit Shock

Actual capital Fixed capital at 2008:Q4

Austria

2004 1206 08 10 2004 1206 08 10 2004 1206 08 10 2004 1206 08 10 2004 1206 08 10

Time

0

50

Ireland

0

50

Switzerland

0

50

Spain

0

50

Germany

India

Turkey

Japan

Finland

Belgium

Canada

France

Australia

United Kingdom

Portugal

Netherlands

Italy

United States

Greece

Sweden

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.Note: Foreign claims of Finland are available after 2010. The graphs are placed in order of difference between 2008:Q4 and 2012:Q4 (ascending).

Figure 13.6 Individual Banking System’s Vulnerability to the Funding Shock



crisis mainly due to higher capital levels. Simulations show that if capital (for all of the banking systems) was held con-stant at the end- of- 2008 levels, then the vulnerability to bank funding flow reversals would have been going up. For the funding shock scenario, there could be two broad groups of countries— vulnerabilities trending down and unchanged (Figure 13.6):

• The European countries in crisis— Ireland, Spain, Por-tugal, Greece— along with some others like the United Kingdom, India, and Turkey have been trending down-ward in their susceptibility to funding shocks. Among these, higher capital buffers seemed to have made a significant difference to India, Turkey, Canada, and the United Kingdom— making these countries more resilient to cross- border funding shocks.

• In Austria, Germany, and Australia, cross- border funding vulnerability is largely unchanged.

There are also fewer propagators of network shocks than before. Comparing the global banking network at the end of 2008 to that at the end of 2012 (Figures 13.7 and 13.8), the number of “arrows” showing the direction of contagion have

dropped. Back in 2008, the United States, the United King-dom, France, and Germany were the main potential propa-gators (leading to at least 10 failures, or half the network) of credit shocks. France, Italy, and Germany were the main contributors to funding shocks. In 2012, the United States and the United Kingdom remained the key potential con-tributors of credit shocks. If the United States and the United Kingdom were to fail, there would be large ripple effects and failures in the rest of the world mainly from their borrowings from the rest of the world. Even though there are no longer major propagators of funding shocks, the United States, the United Kingdom, France, and Germany are still capable of having large impacts on at least two other econo-mies due to funding shocks.2

2 India and Turkey did not fall in the path of ripple effects through fund-ing shocks from the United States, the United Kingdom, France, or Germany in 2012. Banking linkages do not help explain the turmoil in capital flows to India and Turkey experienced during the Federal Re-serve tapering fears in the middle of 2013.

ES

JP

FR

FR

IT

DE

DE

GB

US

GR

NL

SE

IE

CHBE

PTTR AT

US

ES

GB

GR

IE

SECHBE

AU

CA

IN

FI

NL

JP

AU

TR

IT

PT

CAAT IN

1. Credit Shock

2. Funding Shock

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.Note: Blue sphere indicates the banking system that leads to more than 10 (that is, half of the number of countries in the dataset) induced banking failures. Arrows represent how shocks that lead to failure of the banking system are propagated. The figures are constructed with IMF data using the Excel add-in available at nodexl.com. AT = Austria; AU = Australia; BE = Belgium; CA = Canada; CH = Switzerland; DE = Germany; ES = Spain; FI = Finland; FR = France; GB = United Kingdom; GR = Greece; IE = Ireland; IN = India; IT = Italy; JP = Japan; NL = Netherlands; PT = Portugal; SE = Sweden; TR = Turkey; US = United States.

Figure 13.7 Contagion to the Credit Shock and Funding Shock, 2008:Q4


http://nodexl.com


or a crisis from the other country. So, during good times, banking systems can grow and contribute to output growth. However, during stress in other countries, the cross- border credit and funding channels are conduits for bringing home crises from other countries and could have negative GDP growth surprises for the recipient banking system.

Cross- border banking linkages on the credit side do not seem to produce GDP surprises. A panel regression with country fixed effects is estimated to find out whether vulner-abilities to cross- border credit and funding risks explain GDP growth surprises for the 20 countries in the sample (Table 13.2).3 The results show that cross- border credit link-ages and the risks stemming from the linkages seem to be well understood by those making GDP forecasts. While the 2008–09 crisis had negative growth surprises on average for all countries, exposure to credit risk from other banking sys-tems did not significantly make countries better off during normal times, nor did it inflict damage beyond what was expected, during the crisis (Table 13.2, columns 1 and 2).

3 Growth surprise for a country is calculated by actual GDP growth rate minus the forecast of GDP growth rate from Consensus Forecasts.

Are the real effects of cross- border banking linkages well un-derstood by macroeconomic forecasters? In what follows, the chapter tries to gauge whether greater vulnerability to cross- border banking network shocks are already taken into account in the GDP growth forecasts or whether there are major sur-prises. The answer is it depends upon whether the connections are on the assets or the liabilities side of the balance sheet.

5. WHAT IS THE OUTPUT COST OF VULNERABILITY TO BANKING INTERLINKAGES?Extensive cross- border banking linkages bring both benefits and costs. Banking systems can share risk by diversifying their investments across borders so that there is no excessive reliance on good prospects at home. At the same time, banking systems have often relied on foreign funds to sponsor domestic credit growth when times are good or when banks are competing with other banks for market share in a specific loan segment. Both cross- border investments (asset growth) and funding (li-abilities growth) carry the risk of reversal during a global crisis

GBUSIE

FI

IN

DEAU

CA

USGRJP

CA

AT

IE

FR

PT

ITNL

GB AUES

BE

DE

CH

SE

FI

TRIN

CHSE

IT

TR

GR

ES

JP

ATPTBENL

FR

1. Credit Shock

2. Funding Shock

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.Note: Blue sphere indicates the banking system that leads to more than 10 (that is, half of the number of countries in the dataset) induced banking failures. Arrows represent how shocks that lead to failure of the banking system are propagated. The figures are constructed with our data using the Excel add-in available at nodexl.com. AT = Austria; AU = Australia; BE = Belgium; CA = Canada; CH = Switzerland; DE = Germany; ES = Spain; FI = Finland; FR = France; GB = United Kingdom; GR = Greece; IE = Ireland; IN = India; IT = Italy; JP = Japan; NL = Netherlands; PT = Portugal; SE = Sweden; TR = Turkey; US = United States.

Figure 13.8 Contagion to the Credit Shock and Funding Shock, 2012:Q4


http://nodexl.com


As is shown later in this chapter, a random- effects specifi-cation yields an even stronger result for the funding vulner-ability. Every percentage point of potential capital depletion due to higher funding vulnerability increases surprises by 0.03 percentage point during normal times, and reduces sur-prises by 0.06 of a percentage point during crisis, and this effect is economically significant (the null hypothesis for the Wald test is rejected strongly).

Having higher capital buffers of the countries receiving the shocks helps during crises, and has no material impact on real growth surprises during normal times. To see if the measure on network vulnerabilities can be substituted by data on expo-sures and capital separately, a third set of regressions (Ta-ble 13.2, columns 5 and 6) was estimated.6 Results show that higher capital does not lead to lower growth surprises and higher exposures do not contribute to positive growth sur-prises, in general. However, during crises, having higher capi-tal buffers helps to cushion the (negative) surprise impact.

Robustness

The results presented earlier in this chapter are generally ro-bust to different assumptions on parameters for the network analysis and different specifications for the regressions.

6 Financial openness or exposure measured by aggregate statistics (for-eign claims + foreign liabilities)/GDP is a standard regressor in a growth regression.

By contrast, the real effects of possible funding rever-sals due to cross- border interlinkages during crises are not well understood. In good times, countries experience higher growth (surprises) by taking up cross- border fund-ing risks, for instance by extending domestic credit funded from cross- border sources. The estimates (Table 13.2, col-umns 3 and 4) show that during normal (or noncrisis) times, every percentage point potential shortfall in capital levels contributes to a 0.05-percentage-point increase in GDP growth surprise. During crises, however, the benefits could reverse much more, leading to a 0.07-percentage- point decrease in GDP growth surprises over and above the average negative surprises. The same vulnerability re-verses the good outcomes during crisis, although the Wald test on the sum of the coefficients on the funding vulner-ability and the cross term is not always significantly differ-ent from zero.4,5

4 The Wald test on the difference between normal and crisis times cannot reject the null hypothesis (H0: coefficient on funding vulnerability + coefficient on interaction with crisis dummy = 0).

5 A set of regressions with trade linkages was estimated but is not in-cluded in Table 13.2. The trade linkage is measured by (export to and import from the other 19 countries)/GDP. Trade linkages between these countries do not seem to matter for growth surprises during good times or bad times, nor do trade linkages change the outcomes for credit and funding vulnerabilities on growth surprises. This is because trade linkages are typically well documented and included in the dataset while making GDP growth forecasts.

TABLE 13.2

Panel Regression with Country Fixed Effects (Dependent variable: GDP growth rate surprise; sample: 2005–12 [annual, fourth quarter])λ = 0.5, ρ = 0.5 (1) (2) (3) (4) (5) (6)Crisis -3.03*** -3.64*** -3.16*** -1.28 -3.08*** -3.67***

(0.29) (0.63) (0.28) (1.02) (0.28) (0.52)Vul (credit)-1 0.02 0.01

(0.02) (0.02)Vul (credit)-1* Crisis 0.03

(0.03)Vul (funding)-1 0.05** 0.05**

(0.02) (0.02)Vul (funding)-1* Crisis -0.07*

(0.03)Capital-1 -8.12 -7.77

(4.98) (4.96)Capital-1* Crisis 12.39*

(7.13)Exposure-1 0.36 0.38

(0.27) (0.29)Exposure-1* Crisis -0.30

(0.27)Observations 140 140 140 140 140 140R-squared 0.494 0.499 0.513 0.528 0.508 0.521Country pairs 20 20 20 20 20 20

Source: Authors.Note: Standard errors in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01. Capital = capital/GDP; Crisis = 2008–09; Dependent variable = growth rate surprise (actual GDP growth rate - GDP growth rate forecast) in percentage points; Exposure = (foreign claims + foreign liabilities)/GDP; Vul (.) = the vulnerability index from the network analysis.



• The result that higher capital buffers help cushion negative output surprises during crisis is robust to different model specifications and different data on capital from the IMF Financial Soundness Indica-tors database, where the data start in 2008, instead of Bankscope.

6. CONCLUSIONSTo summarize, banking systems’ vulnerabilities from cross- border network linkages were found to have decreased in the last five years. For both asset- and liability-side vulnera-bilities, on average for the global banking system, the po-tential for capital depletion arising from credit risks and funding risks has come down since the global financial cri-sis. The reduction is mainly due to lower exposures, but is also partly due to higher capital buffers around the world.

While the trend is similar for individual countries, the reason for the decline in vulnerabilities differs between countries and between credit and funding for particular countries. It was also found that, compared to 2008, the

• Indices constructed with different lambda and rho: The movement of the indices is similar to the original indices if different parameters are used. The initial vulnerability measures are highly correlated (above 0.9) to the new indices constructed with different parameters. The trends in these indices are similar between various assumptions on the parameters for their construction: λ and ρ (Figures 13.9 and 13.10).

• In the regression part, the findings regarding the fund-ing and credit vulnerability indices are robust to various assumptions on the parameter values (λ and ρ) for the network analysis. The cross- product terms (crisis * vul-nerability) are also still significant for most parameter values. Table 13.3 shows one such set of parameters.

• Rerunning the regressions using random, instead of fixed, effects gives a stronger result on the funding risk (Table 13.4). As mentioned earlier in this chap-ter, higher funding vulnerability significantly exac-erbates negative output surprises. In general, results of panel regressions with random effect are overall similar to the baseline result.

0.30.40.50.60.7

10

0

40

20

30

062004 08 10 12

Inde

x of

vul

nera

bilit

y

Time

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.

Figure 13.9 Credit Vulnerability Indices—Varying λ(Weighted average of vulnerability with different parameters; global banking system, 2005–12)

0.30.40.50.60.7

10

0

40

20

30

062004 08 10 12

Inde

x of

vul

nera

bilit

y

Time

Sources: Bank for International Settlements; Bankscope; and authors’ estimates.

Figure 13.10 Funding Vulnerability Indices— Varying ρ(Weighted average of vulnerability with different parameters; global banking system, 2005–12)



border credit. Therefore, risks from cross- border borrowing need much more analysis and understanding than just looking at overall external funding volumes. In particular, taking on higher funding risks (by borrowing more from cross- border sources) generally exacerbates the negative output sur-prise during crisis. This finding is robust to different values of the parameters used to create the vulnerability indices and different specifications and estimation methods of the regres-sion model.

Regardless of network effects, higher capital helps dur-ing a crisis, and it does not hurt to raise it during normal times. Higher capital buffers help mitigate negative GDP surprises during crisis, but the same buffers might not have a real impact during normal times. These findings give ad-ditional reasons for strengthening buffers during normal times, since it does not seem to have a significant impact on output surprises.

Future research could try to explain why funding risks appear to matter more than credit risks. One reason could be the transparency of credit links apparent with the published BIS data and a general understanding of the cross- border credit exposures of banks from certain countries. For in-stance, it is well known that the Spanish and Austrian banks have large credit exposures in Latin America and central and eastern Europe, respectively. However, there is less docu-mentation about which countries Spanish and Austrian banks (and other sectors) borrow from. The BIS Consoli-dated Statistics do not provide liability- side information. As mentioned before, such information was only derived by making assumptions. Policymakers need to understand the specific vulnerabilities from funding linkages while making macroeconomic forecasts, and this chapter has made the case for the need to access better data.

REFERENCESAlbertazzi, U., and M. Bottero. 2014. “Foreign Bank Lending:

Evidence from the Global Financial Crisis.” Journal of Interna-tional Economics 92, Supplement 1 (0): S22–S35.

Cerutti, Eugenio. 2013. “Banks’ Foreign Credit Exposures and Borrowers’ Rollover Risks Measurement, Evolution and Deter-minants.” IMF Working Paper 13/9, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/ Banks- Foreign- Credit- Exposures - and- Borrowers- Rollover- Risks-Measurement-Evolution-and -40235.

Cerutti, Eugenio, Stijn Claessens, and Patrick McGuire. 2011. “Systemic Risk in Global Banking: What Available Data Can Tell Us and What More Data Are Needed.” IMF Working Paper 11/222, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31 / Systemic- Risks- in- Global- Banking- What- Available- Data- Can - Tell- Us- and-What-More-Data-Are-25259.

Cetorelli, N., and L. S. Goldberg. 2010. “Global Banks and Inter-national Shock Transmission: Evidence from the Crisis.” IMF Economic Review 59 (1): 41–76.

Cihák, Martin, Sonia Munoz, and Ryan Scuzzarella. 2011. “The Bright and the Dark Side of Cross- Border Banking Linkages.” IMF Working Paper 1/41, International Monetary Fund,

number of countries as core propagators of credit and fund-ing shocks have dropped. The United Kingdom and the United States would still be the major propagators of credit shocks in 2012.

Funding risks have significant positive effects on growth surprises during normal times and significant negative ef-fects on growth surprises during crisis times. Risks from cross- border borrowing have significant impacts on real growth surprise and these risks are higher than those from cross-

TABLE 13.3

Robustness: Panel Regression with Country Fixed Effects(Dependent variable: GDP growth rate surprise; sample: 2005–12 [annual, fourth, quarter])λ = 0.3, ρ = 0.3 (1) (2) (3) (4)Crisis -3.07*** -3.44*** -3.17*** -2.20***

(0.29) (0.54) (0.28) (1.02)Vul (credit)-1 0.04 0.03


(0.03)Vul (funding)-1 0.11*** 0.12**

(0.04) (0.02)Vul (funding)-1* Crisis -0.07*

(0.03)Observations 140 140 140 140R-squared 0.500 0.502 0.519 0.535Country-pairs 20 20 20 20

Source: Authors.Note: Standard errors in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01. Capi-tal = capital/GDP; Crisis = 2008–09; Dependent variable = growth rate surprise (actual GDP growth rate - GDP growth rate forecast) in percent-age points; Exposure = (foreign claims + foreign liabilities)/GDP; Vul (.) = the vulnerability index from the network analysis.

TABLE 13.4

Robustness: Panel Regression with Random Effects(Dependent variable: GDP growth rate surprise; sample: 2005–12 [annual, fourth quarter])λ = 0.5, ρ = 0.5 (1) (2) (3) (4)Crisis -3.04*** -3.65*** -3.00*** -0.48***

(0.28) (0.61) (0.28) (0.96)Vul (credit)-1 0.02 0.01


(0.03)Vul (funding)-1 0.01 0.03*

(0.01) (0.01)Vul (funding)-1* Crisis -0.09***

(0.03)

Observations 140 140 140 140R-squared 0.453 0.458 0.441 0.473Country-pairs 20 20 20 20

Source: Authors.Note: Standard errors in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01. Capital = capital/GDP; Crisis = 2008–09; Dependent variable = growth rate surprise (actual GDP growth rate - GDP growth rate forecast) in percentage points; Exposure = (foreign claims + foreign liabilities)/GDP; Vul (.) = the vulnerability index from the network analysis.





https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Banks-Foreign-Credit-Exposures-and-Borrowers-Rollover-Risks-Measurement-Evolution-and-40235





———. 2013. World Economic Outlook— Transitions and Tensions, Chapter 3. Washington, DC, October. https://www.imf.org /external/pubs/ft/weo/2013/02/.

Kalemli- Ozcan, Sebnem, Elias Papaioannou, and Fabrizio Perri. 2013. “Global Banks and Crisis Transmission.” Journal of Inter-national Economics 89 (2): 495–510.

Kim, Kyunghun, and Srobona Mitra. 2014. “Real and Financial Vulnerabilities from Cross- Border Banking Linkages.” IMF Working Paper 14/136, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/ Real- and- Financial- Vulnerabilities-from-Crossborder -Banking-Linkages-41792.

Minoiu, Camelia, Chanhyun Kang, V.S. Subrahmanian, and Ana-maria Berea. 2015. “Does Financial Connectedness Predict Cri-ses?” Quantitative Finance 15 (4): 607–24.

Nier, E., J. Yang, T. Yorulmazer, and A. Alentorn. 2007. “Network Models and Financial Stability.” Journal of Economic Dynamics and Control 31 (6): 2033–60.

Popov, A., and G. F. Udell. 2012. “ Cross- Border Banking, Credit Access, and the Financial Crisis.” Journal of International Eco-nomics 87 (1): 147–61.

Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ The- Bright- and- the- Dark- Side- of-Cross -Border-Banking-Linkages-25147.

De Haas, R., and I. Van Lelyveld. 2014. “Multinational Banks and the Global Financial Crisis: Weathering the Perfect Storm?” Journal of Money, Credit, and Banking 46 (s1): 333–64.

Espinosa- Vega, Marco, and Juan Solé. 2011. “ Cross- Border Finan-cial Surveillance: A Network Perspective.” Journal of Financial Economic Policy 3 (3): 82−205.

Giannetti, M., and L. Laeven. 2012. “The Flight Home Effect: Evidence from the Syndicated Loan Market during Financial Crises.” Journal of Financial Economics 104 (1): 23–43.

International Monetary Fund (IMF). 2009. Global Financial Sta-bility Report— Responding to the Financial Crisis and Measuring Systemic Risks, Chapter 2. Washington, DC, April. https://www.imf.org/external/pubs/ft/gfsr/2009/01/.

———. 2010. Global Financial Stability Report— Meeting New Challenges to Stability and Building a Safer System, Chapter 2. Washington, DC, April. https://www.imf.org/en/Publications /GFSR/Issues/2016/12/31/ Meeting- New- Challenges- to -Stability-and-Building-a-Safer-System.


https://www.imf.org/external/pubs/ft/weo/2013/02/

https://www.imf.org/external/pubs/ft/weo/2013/02/











CHAPTER 14

Credibility and Crisis Stress Testing

LI LIAN ONG • CEYLA PAZARBASIOGLU

Credibility is the bedrock of any crisis stress test. The use of stress tests to manage systemic risk was introduced by the US authorities in 2009 in the form of the Supervisory Capital Assessment Program. Supervisory authorities in other jurisdictions conducted similar exercises in the fol-

lowing months. In some of those cases, the design and implementation of certain elements of the framework were criticized for their lack of credi-bility. This chapter proposes a set of guidelines for constructing an effective crisis stress test. It combines financial markets impact studies of previous exercises with relevant case study information gleaned from those experiences to identify the key elements and to formulate their ap-propriate design. Pertinent concepts, issues, and nuances particular to crisis stress testing are also discussed. The findings may be useful for country authorities seeking to include stress tests in their crisis management arsenal as well as for the design of crisis programs.

“Investors don’t like uncertainty. When there’s uncertainty, they always think there’s another shoe to fall.”Kenneth Lay, then- CEO of Enron Corp.

August 20, 2001

the availability of a financial backstop, and the subsequent publication of the methodology and results appeared to reas-sure markets (see Peristiani, Morgan, and Savino 2010).

The SCAP represented a high- profile adoption of forward- looking techniques of stress testing that assessed the preparedness of the banks (and authorities) to deal with low- probability, high- impact events. The subsequent findings re-vealed that the capital needs of the largest US banks at the time would be manageable even if a more adverse scenario were to materialize (see Tarullo 2010). Investor sentiment rebounded and stabilized, and the assessed banks were able to add more than $200 billion in common equity in the fol-lowing 12 months. The US supervisors have since followed up on the SCAP with regular, publicized supervisory stress tests, the Dodd- Frank Act Stress Test, which has become a separate part of the Comprehensive Capital Analysis and Review (CCAR) (Federal Reserve 2018a, 2018b).

Crisis stress tests conducted by supervisors should not be confused with supervisory stress tests undertaken during cri-ses. Both types of stress tests may be used for similar purposes, that is, to: (1) determine a needed capital buffer over current

1. INTRODUCTIONStress tests have become the “new normal” in financial crisis management. They were used by country authorities as an instrument for regaining the public’s trust in the banking system during the global financial crisis. This new tool, known as a “crisis stress test,” is essentially a supervisory ex-ercise accompanied by detailed public disclosure to remove widespread uncertainty about banks’ balance sheets and the authorities’ plans for those banks. Put another way, the crisis stress test is a microprudential exercise with macropruden-tial objectives (Figure 14.1). In this regard, transparency, and hence the quality of disclosure, is critical.

The concept of crisis stress testing was introduced by the US authorities in early 2009 in the form of the Supervisory Capital Assessment Program (SCAP). The solvency stress test-ing exercise took place during the darkest days of the sub-prime loans meltdown, following a sharp loss of confidence in US banks and an unprecedented decimation of their market value. The announcement of the SCAP itself was initially met with trepidation and skepticism by markets, but official clari-fications surrounding the event about the aim of the exercise,

This chapter was previously published as an article in the International Journal of Financial Studies (Ong and Pazarbasioglu 2014). The authors would like to thank Joaquin Gutierrez for his invaluable input; and Francesco Columba, Piers Haben, Daniel Hardy, Heiko Hesse, Emanuel Kopp, Toshitake Kurosawa, Caroline Liesegang, Lusine Lusinyan, Dermot Monaghan, Marina Moretti, Mohamed Norat, Antonio Pancorbo, Alvaro Piris, Til Schuermann, Liliana Schumacher, the Irish and Spanish authorities, and three anonymous referees for their useful comments. All remaining mistakes are the responsibility of the authors.


Credibility and C

risis Stress Testing314

Source: Jobst, Ong, and Schmieder 2013.Note: The IMF staff typically defines top- down stress tests as those that are either conducted using the data of individual banks and then aggregated, or on an aggregated portfolio; bottom- up stress tests are defined as those conducted by individual institutions using their own internal risk models and data. The IMF staff had previously conducted a rudimentary stress test of the Hungarian banking system during the crisis program discussions in late 2008 as an input into determining the size of the program, which were subsequently published (IMF 2008). CCAR = Comprehensive Capital Analysis and Review; CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; FSAPs = Financial Sector Assessment Programs; SCAP = Supervisory Capital Assessment Program.

Figure 14.1 Solvency Stress Testing Applications

Bank SolvencyStress Testing

Macroprudential

Surveillance CrisisManagement

Microprudential

Supervisory

Top Down(for example, CCAR)

Bottom Up(for example, CCAR)

Top Down(for example, SCAP,

CEBS/EBA,Spain, IMF crisis

programs)

Bottom Up(for example, SCAP,

CEBS/EBA)

Top Down(for example, FSAPs,GFSR, central bankfinancial stability

units)

Bottom Up(for example, FSAPs)

RiskManagement

Internal RiskManagement

Bottom Up(for example,

financialinstitutions’ own)


Li Lian Ong and Ceyla Pazarbasioglu 315

public confidence. Indeed, this aspect of regaining the public’s trust in the financial system is at least as important as, if not more important than, just shedding light on the conditions of banks themselves. The ensuing discussion shows how percep-tions of the credibility of supervisory and regulatory authorities could influence the design of key aspects of a crisis stress test.

The assessment by the Turkish authorities of its banking sector following the 2001 crisis is an example of a public di-agnostic exercise that helped to restore confidence in the banking sector and its supervisor and regulator. Although it did not include a forward- looking stress test component, its overall objective and design included the necessary attri-butes for a credible outcome. The financial status of all do-mestic banks was assessed using improved accounting standards and a three- stage audit procedure, the first two of which were by independent auditors. The capital adequacy of each bank was determined and banks that were under-capitalized were required to take capital action. A financial backstop through the State Recapitalization Scheme was made available to banks that were deemed solvent, but which were unable to raise the necessary capital. The objective, method, and implementation details of the exercise were published (see Banking Regulation and Supervision Agency 2002a), as were the findings and progress on actions taken (Banking Regulation and Supervision Agency 2002b).

This chapter focuses on the design of crisis stress tests, leaving the comprehensive study of other aspects of a diag-nostic to future research. Work on developing a comprehen-sive framework for an effective crisis stress test has been limited to date. Hirtle, Schuermann, and Stiroh 2009 draw lessons from the SCAP in analyzing the complementarities between macroprudential and microprudential supervision. Schuermann 2012 explores in some detail the design of stress scenarios and their application in terms of modeling losses, revenues, and balance sheets— key elements in macro stress testing— in the US, EU, and Ireland exercises. He also exam-ines the disclosure strategies across the various exercises. Else-where, Langley 2013 assesses more broadly the “anticipatory techniques” applied in the SCAP and their performative power vis- à- vis those applied in the European exercises. Other empirical and policy- related literature in this area has largely focused on the effectiveness of the SCAP (Bernanke 2010; Matsakh, Altintas, and Callender 2010; Peristiani, Morgan, and Savino 2010; Tarullo 2010), with some coverage of the European stress tests (Onado and Resti 2011).

The specific objective of this chapter is to formulate guide-lines for designing a crisis solvency stress test, based on lessons learned from previous experiences. Although a crisis stress testing exercise may cover either solvency or liquidity risk or both, this chapter focuses on the former. In this regard, the study complements the work done by Hirtle, Schuermann, and Stiroh 2009, and Schuermann 2012. Various methodol-ogies are employed in the analysis:

• First, the study distinguishes the effective crisis stress tests using financial market impact studies of exer-cises in the United States, the European Union, Ire-land, and Spain, including analyzing the statistical

solvency levels; (2) differentiate the soundness of banks in the system as part of triage analysis; and/or (3) quantify potential fiscal costs, depending on the magnitude of the projected cap-ital shortfalls and the urgency of any required recapitalization. However, supervisory stress tests in crisis situations may be different from crisis stress tests in that: (1) the focus of the former may be to assess the condition of individual banks solely for microprudential rather than for macroprudential or system- wide stability purposes (see IMF 2012a); and (2) the former typically do not have the same degree of (public) trans-parency and indeed, may have to be kept confidential to avoid potentially unleashing an unmanageable backlash if the key elements necessary for publication— which will be discussed in much of this chapter— are not in place.

Clearly, crisis stress tests must be credible to be success-ful. As in the United States, supervisory authorities in Eu-rope have used crisis stress tests for systemic risk management but with varying degrees of effectiveness. This suggests that the design of such exercises matters significantly, notably:

• The governance of the tests (that is, the stress tester and the overseer) must be perceived to be indepen-dent, with the requisite technical expertise.

• The stress tests themselves must be sufficiently strin-gent, yet plausible. The scope, coverage, scenario design, and methodology need to be considered sufficiently comprehensive and robust to capture key risks to the institutions and system.

• The stress tests should be simultaneous, consistent, and comparable cross- firm assessments to enable a broader analysis of risks and an evaluation of estimates for indi-vidual institutions (Tarullo 2010). From a macropru-dential perspective, they should allow for a better understanding of interrelationships across institutions.

• The stress tests should usefully inform markets about the risks associated with the banks, and the results must be sufficiently granular such that there is clear differentiation among institutions in the first in-stance, to guide subsequent actions.

• Last, but not least, the manner in which the stress test results will be backstopped or used must be clar-ified early on to guide depositors and investors.

Crisis stress tests should be seen as one element of an over-all strategy to rebuild public confidence in a banking system. Ideally, such a strategy should include (1) containment; (2) diagnostics (asset quality review [AQR], data integrity and verification [DIV], stress test); and (3) restructuring or exit. Within the diagnostics component, the stress test itself is a forward- looking tool for determining a capital buffer against further deterioration in the real economy. As is discussed later in this chapter, the preceding AQR and DIV of banks’ portfolios are critical for credibility, as they help to ensure that the data used in the stress test are “clean.” However, the nature and extent of these exercises may differ, depending on market perceptions of the reliability of the reported informa-tion and the design of the stress test.

In some cases, the restoration of the credibility of financial supervisors and regulators is another element in rebuilding


Credibility and Crisis Stress Testing316

and the fiscal purse. The study’s findings may be useful for authorities seeking to undertake stress tests for systemic risk management and for the design of financial crisis programs.

The chapter is organized as follows. Section 2 details the relevant case studies of stress testing exercises conducted in the United States and Europe, as well as the market data used in the initial impact study. Section 3 discusses the met-rics used for defining the effectiveness of those crisis stress tests and presents the empirical analysis. Section 4 draws on those findings and the qualitative information gleaned from the case studies to identify the key stress test elements and to formulate their design. A sidebar comparing the differences between bank restructuring costs and loss estimates from crisis stress tests is presented in Section 5. Section 6 concludes.

2. THE DATAThe analysis draws on four case studies covering seven crisis stress tests. The tests were conducted in the United States, the European Union, Ireland, and Spain between 2009 and 2012 (Table 14.1). The details of the individual exercises are sourced from the respective authorities, namely, the Board of Governors of the Federal Reserve, the European Banking Association (EBA) and its predecessor, the Committee of European Banking Supervisors (CEBS), the Central Bank of Ireland (CBI), and Banco de España (BdE).

The CEBS had noted that its stress tests contrasted with the crisis stress test nature of the SCAP. The EU authority had stated that the objective of its exercise was to “provide policy information for the assessment by individual Member States of the resilience of the EU banking sector as a whole and of the banks participating in the exercise,” compared to the SCAP, which

performance of the respective financials stock indi-ces and sovereign credit default swap (CDS) spreads around the announcement of the stress test results and in the months following those announcements.

• Next, a case study analysis is applied to identify the key elements of a crisis stress test and to formulate the appropriate design of those elements, drawing on qualitative information from previous stress tests.

• Where relevant, the analysis is juxtaposed against some of the relevant “best practice” principles pre-sented in the literature (for example, Basel Commit-tee for Banking Supervision 2008; Federal Reserve, Federal Deposit Insurance Corporation, and Office of the Comptroller of the Currency 2012; IMF 2012a), while highlighting concepts, issues, and nu-ances that may be particular to crisis stress testing.

The study’s conclusions point to an immutable fact, which is that crisis stress tests are not for the half- hearted. Ideally, the stress test should take place sufficiently early to address any crisis of confidence in the banking system and have a clearly specified objective. Moreover, lessons learned from past experiences show that country authorities must be fully committed if they are to undertake such an exercise, lest it backfire. The authorities must be prepared to conduct a thorough, honest, and transparent examination of their banking system and resolve to take appropriate follow- up action(s) on the results with the necessary resources to back them, if the exercise is to serve its purpose. Supporting ac-tivities such as AQRs and possibly follow- up stress tests are necessary to ensure the credibility of crisis stress tests. How-ever, political economy considerations could also play an im-portant role in the design of crisis stress tests, given the potential implications of the results for public confidence

TABLE 14.1

Case Studies: Crisis Stress TestsJurisdiction Stress Testing Exercise Stress Tester Participating AuthoritiesUnited States Supervisory Capital Assessment

Program (SCAP) 2009Authorities Federal Reserve, Federal Deposit Insurance

Corporation, Office of the Comptroller of Currency

European Union Committee of European Banking Supervisors (CEBS) 2009

Authorities National supervisory authorities, CEBS, European Commission (EC), and European Central Bank (ECB)

Committee of European Banking Supervisors 2010

Authorities National supervisory authorities, CEBS, EC, and ECB

European Banking Authority (EBA) 2011

Authorities National supervisory authorities, EBA, EC, ECB, and European Systemic Risk Board

Ireland Prudential Capital Assessment Review (PCAR) 2011

Authorities with loan loss inputs from BlackRock Solutions

Central Bank of Ireland

Spain Top Down 2012 Oliver Wyman and Roland Berger

Banco de España (BdE), Ministry of Economy and Competitiveness, the Troika, and representatives from two EU Member States

Bottom Up 2012 Oliver Wyman BdE, Ministry of Economy and Competi-tiveness, the Troika, and EBA

Sources: BdE; Central Bank of Ireland; EBA; and Federal Reserve.



was linked to “determining the individual capital needs of banks” (CEBS 2010). However, the CEBS’s overt efforts at trans-parency to reassure markets— including through the announce-ment of aggregate results in the 2009 exercise— were consistent with the macroprudential application of crisis stress tests.

The chapter first identifies successful crisis exercises by an-alyzing the performance of market indicators, consistent with existing studies. Previous research had examined the behavior of stock prices of individual US banks post- SCAP (Matsakh, Altintas, and Callender 2010; Peristiani, Morgan, and Savino 2010) as well as the sovereign CDS spreads (Peristiani, Mor-gan, and Savino 2010; Schuermann 2012) to determine the effectiveness of the respective crisis stress tests. Here:

• The financials stock price indices for each jurisdiction are studied as proxies for the market’s assessment of the soundness of the respective banking systems. Stock prices represent a bellwether indicator for mar-ket confidence in that shareholders are the “first loss” investors and the evidence shows that they respond very quickly to incorporate all relevant publicly avail-able information in their pricing (Fama 1970).

• The behavior of sovereign CDS spreads around the stress testing exercises and related events is also con-sidered. Sovereign CDS spreads provide an indica-tion of the perceived creditworthiness of a country, which is considered closely linked to the health of its banking sector, given the potential implications for the public purse if government support is required (Mody and Sandri 2011). In several banking sys-tems, the high holdings of sovereign debt have fo-cused market concerns on the bank- sovereign feedback loop (Acharya, Drechsler, and Schanabl 2011; Committee on the Global Financial System 2011; Angeloni and Wolff 2012; Darracq Paries, Faia, and Palenzuela 2013).

All market data used in this study are sourced from Bloom-berg (Table14.2). It should also be noted that caveats apply to the use of financial market indicators to define the ef-fectiveness of the stress tests in that they may also be influ-enced by other concurrent events, which are not isolated in this study.

3. IDENTIFYING THE SUCCESSFUL CRISIS STRESS TESTSA simple event study- type methodology is applied for determining the effectiveness of a crisis stress test. Given that the analytical framework is not strictly that of a formal event study, the assessment shall be referred to as an “impact study.” A stress test is classified as successful if it has been able to stabilize or improve investor sentiment toward the banking system for at least six months after the results are announced, providing sufficient time for follow- up action(s) to be taken. In other words, the stress test is considered to have achieved its objective if it has been able to establish a “floor” for the market during this period (Table 14.3), which is assessed based on the following “expectations” in the six months fol-lowing the announcement of the results:

• The return on the financials stock index is relatively stable or rises.

• The volatility of daily returns (calculated as the stan-dard deviation over 130 days) stabilizes or declines (relative to the preceding six months).

• The sovereign CDS spread stabilizes or narrows.In this context, the empirical evidence from the US

SCAP shows that the exercise had been successful in achiev-ing its aim (Figure 14.2):

• The release of the SCAP results effectively halted and then reversed the two- year slide in investor confi-dence toward the country’s banks. The financials index rose by almost 20 percent in the following six months. At the same time, market volatility— which had peaked just prior to the exercise— declined sharply over this period. The S&P 500 Financials Sector Index subsequently remained above the level estab-lished by the SCAP results, although it flirted with that floor during the more volatile period in the third quarter of 2012.

• US firms substantially increased their capital follow-ing the SCAP. The weighted Tier 1 (T1) Common Equity ratio of the 18 bank holding companies that were in the SCAP sample more than doubled from an average 5.6 percent at the end of 2008 to 11.3

TABLE 14.2

Market Data: Financials Stock Price Index and CDS Spreads

Jurisdiction

Stock Market Credit Default Swaps

Proxy Index Bloomberg Ticker Proxy Index Bloomberg TickerUnited States S&P 500 Financials Sector

IndexS5FINL [Index] United States EUR senior

5-yearZCTO CDS EUR SR 5Y [Corp]

European Union STOXX Europe 600 Banks Price EUR

SX7P [Index] iTraxx SovX Western Europe USD 5-year

ITRX SOVX WE CDSI GEN 5Y [Corp]

Ireland Irish Stock Exchange Financial Index

ISEF [Index] Ireland USD senior 5-year IRELND CDS USD SR 5Y [Corp]

Spain MSCI Spain Financials Index

MSES0FN [Index] Spain USD senior 5-year SPAIN CDS USD SR 5Y [Corp]

Source: Bloomberg.Note: CDS = credit default swap.


Credibility and C


TABLE 14.3

Crisis (and Follow-Up) Stress Tests: Performance Statistics1

Indicator

Effectiveness of Stress Test

United States European Union Ireland Spain

Crisis Supervisory Crisis Crisis Surveillance Crisis

SCAP 2009 CCAR 2011 CCAR 2012 CCAR & DFA 2013

CEBS 2009 CEBS 2010 EBA 2011 PCAR 2011 & IMF2

FSAP 20123 TD 2012 BU 2012

Instrument Measure Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post Pre PostFinancials

stock indexIndex return

(160-day, in percent)

-13.7 19.1 11.4 -20.5 24.0 1.5 17.3 8.9 63.4 -1.7 -0.3 2.3 -19.5 -21.7 -67.2 78.9 -24.4 21.8 -21.4 23.8 -1.8 -5.6

Return volatility (160-day, in percent)

6.4 2.5 1.2 2.1 2.2 1.2 0.9 0.9 2.1 1.6 2.2 1.4 1.2 2.9 4.1 3.7 2.2 2.5 2.2 2.5 2.8 1.7

Credit default swap

Spread (160-day, in basis points)

-6 -7 -4 8 -19 4 2 -14 n.a. 33 43 52 104 67 204 -193 161 -284 184 -288 −44 -90

Sources: Bloomberg; and authors’ calculations.Note: BU = bottom up; CCAR = Comprehensive Capital Analysis and Review; CEBS = Committee of European Banking Supervisors; DFA = Dodd-Frank Wall Street Reform and Consumer Protection Act; EBA = European Banking Authority; FSAP = Financial Sector Assessment Program; PCAR = Prudential Capital Assessment Review; SCAP = Supervisory Capital Assessment Program; TD = top down.1Relative to announcement of stress test results.2The publication of the IMF’s Third Review in September 2011 indicating that the outcomes of the PCAR were being incorporated into banks’ recapitalization and restructuring plans helped provide credibility to the exercise. 3Included for completeness only—not intended as a crisis stress test; surveillance stress testing exercise was conducted in a crisis environment.



Sour

ces:

Blo

omb

erg;

Fed

eral

Res

erve

; var

ious

fina

ncia

l med

ia; a

nd a

utho

rs’ c

alcu

latio

ns.

Not

e: C

AP

= C

apita

l Ass

ista

nce

Prog

ram

; CC

AR

= C

omp

rehe

nsiv

e C

apita

l Ana

lysi

s an

d Re

view

; CD

S =

cre

dit d

efau

lt s

wap

; LH

S =

left

- han

d si

de ;

RHS

= ri

ght-

hand

sid

e; S

CA

P =

Sup

ervi

sory

C

apita

l Ass

essm

ent P

rogr

am; T

ARP

= T

roub

led

Ass

et R

elie

f Pro

gram

.

Fig

ure

14.

2 U

nite

d St

ates

: The

Sen

timen

t aft

er th

e SC

AP

(In

dexe

d to

100

on

Febr

uary

20,

200

7)

Stoc

k re

turn

vol

atili

ty(R

HS)

Even

tS&

P 50

0 Fi

nanc

ial S

ecto

r Ind

ex(L

HS)

US C

DS E

UR s

enio

r five

-yea

r(In

verte

d, L

HS)

SCAP

resu

lts s

tock

inde

x le

vel

0

120 2080 4060100

08 15 237 46

Jan.

200

7Ja

n. 1

2Ja

n. 1

1Ja

n. 1

0Ja

n. 0

9Ja

n. 0

8Ja

n. 1

3

Feb.

9, 2

009:

SCA

P ex

erci

se a

nnou

nced

Feb.

10,

200

9: C

AP a

nnou

nced

Feb.

23–

25, 2

009:

CAP

cla

rified

Feb.

24,

200

9: S

CAP

desi

gn a

nd im

plem

enta

tion

publ

ishe

d; C

DS s

prea

d w

ides

t

May

7: 2

009

SCAP

resu

lts a

nnou

nced

Oct.

7, 2

009:

US

budg

et d

efici

tnu

mbe

r ann

ounc

ed b

y CB

O

Mar

ch 7

, 201

3: D

FA 2

012/

13st

ress

test

resu

lts a

nnou

nced

Mar

ch 1

3, 2

013:

CCA

R 20

12/1

3re

sults

ann

ounc

ed

Oct.

1, 2

013:

US

gove

rnm

ent

shut

dow

n ov

er d

ebt c

eilin

g

Feb.

20,

200

7: F

inan

cial

sst

ock

inde

x pe

akOc

t. 13

, 200

8:TA

RP a

nnou

nced

Mar

ch 6

, 200

9: F

inan

cial

sst

ock

inde

x tro

ugh

Mar

ch 1

8, 2

011:

CCA

R 20

10/1

1ov

ervi

ew p

ublis

hed

Mar

ch 1

3, 2

012:

CCA

R 20

11/1

2re

sults

ann

ounc

ed



2011a). The announcement of the Outright Mon-etary Transactions by the ECB in August 2012 further improved market confidence in the region as a whole.

• In Ireland, the Prudential Capital Assessment Re-view (PCAR 2011) contributed to stabilizing market sentiment. However, it was not until the publication of the IMF’s Third Review in September 2011—six months after the release of the PCAR results— indicating that the program’s structural benchmarks had largely been met and that the outcomes of the PCAR were being incorporated into banks’ recapital-ization and restructuring plans (IMF 2011), that the exercise gained credibility. In the following six months, the financials stock price index rose by almost 80 percent— albeit from a very low base— the volatility of returns fell, and the sovereign CDS spreads tight-ened by more than 190 basis points (Figure 14.4).

• In Spain, the third- party bottom- up stress test and corresponding revelation of a comprehensive strategy to identify and deal with problem banks stabilized market sentiment. The announcement of the IMF Fi-nancial Sector Assessment Program and third- party top- down stress test results coincided with increased volatility in stock price returns, but also signaled that the authorities were closer to taking concerted ac-tion to restructure the banking sector (IMF 2012b; Roland Berger 2012; Oliver Wyman 2012a). The sub-sequent publication of the memorandum of under-standing with the Eurogroup in July 2012, which incorporated comprehensive diagnostics of banks’ balance sheets and the details of a financial backstop, reassured investors. Stock price volatility declined sharply and sovereign CDS spreads narrowed by 90 basis points in the six- month period following the re-lease of the bottom- up results (Figure 14.5).

A summary of the effectiveness of the respective crisis stress tests is presented in Table 14.4.

4. DESIGNING AN EFFECTIVE CRISIS STRESS TESTCrisis stress tests require additional considerations that may not be required of supervisory stress tests during normal times. In particular, the design of certain elements may necessarily be different from what is typically done in the latter (for example, the timing of the test; its governance; the transparency requirements; and the objective, action plan, and financial backstop). Other aspects have to be constructed to withstand intense public scrutiny (for example, the scope and scenario design). While no one particular element can alone ensure the success of a crisis stress test, each one plays a crucial part in the credibility of the exercise as a whole.

There are also additional activities that provide integral support for or complement crisis (solvency) stress tests. These

percent in the fourth quarter of 2012, reflecting an increase in T1 Common Equity from $393 billion to $792 billion during the same period.

• US CDS spreads narrowed in tandem with the im-provement in the financials index during the SCAP period. However, they subsequently dissociated from developments in the banking sector in Septem-ber 2011 as markets turned their attention to the fis-cal deficit after the Congressional Budget Office announced that the US budget deficit had reached its widest as a percentage of GDP since 1945.

The turnaround in confidence in US banks from the SCAP buoyed sentiment toward banks elsewhere, at least temporarily. EU banks’ stock prices benefited from the re-bound and volatility fell; however, they were unable to sus-tain the gains over the medium term, with some countries having to conduct separate tests subsequently:

• The stress tests of EU banking systems were less con-vincing. Although stock prices remained relatively stable following the announcements of the CEBS 2009 and 2010 results, the sovereign CDS spreads continued to widen (Figure 14.3):– The CEBS 2010 exercise suffered the ignominy

of having Ireland request a bailout from the European Commission (EC), the European Cen-tral Bank (ECB), and the IMF (“the Troika”) weeks after the stress tests indicated that EU banks would remain sufficiently capitalized and resilient under adverse scenarios (CEBS 2009, 2010a). The financials stock index followed a downward trend, and despite a turnaround, had not returned to the levels recorded around the time of the CEBS 2009 stress test some four years later.

– Similarly, systemically important banks Dexia (Belgium) and Bankia (Spain) passed the EBA 2011 stress test (EBA 2011b) only to require significant restructuring within a few months. These events were accompanied by sharp jumps in the volatil-ity of stock market returns.

– The EU Capital Exercise (EBA 2011e) was subse-quently announced in October 2011, in response to a rapidly evolving crisis. The disclosure of its results in December 2011, followed by the intro-duction of the ECB’s Long- Term Refinancing Operation (LTRO) facility for financing euro area banks later that month, halted the deteriora-tion in confidence toward EU sovereigns as evi-denced by their narrowing CDS spreads. In the former, the EBA reviewed banks’ actual capital positions as at the end of June 2011 and their sov-ereign exposures in light of the worsening of the sovereign debt crisis in Europe, and requested that they set aside additional capital buffers by June 2012 based on their respective September 2011 sover-eign exposure figures and capital positions (EBA


Li Lian Ong and C

eyla Pazarbasioglu321

Sources: Bloomberg; EBA; various financial media; and authors’ calculations.Note: CDS = credit default swap; CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; ECB = European Central Bank; EU = European Union; LTRO = long- term refinancing operation; LHS = left- hand side; OMT = outright monetary transactions; RHS = right- hand side.

Figure 14.3 European Union: The Ebb from the EBA (Indexed to 100 on April 20, 2007)

Stock return volatility(RHS)

Event STOXX Europe 600 BanksPrice Index (LHS)

iTraxx SovX Western EuropeUSD five-year (inverted, LHS)

CEBS 2009 resultsstock index level

0

120

20

80

40

60

100

0

8

1

5

2

3

7

4

6

Jan. 2007 Jan. 12Jan. 11Jan. 10Jan. 09Jan. 08 Jan. 13

May 12, 2009: CEBS 2009exercise announced

Oct. 10, 2011: Dexianationalized and dismantled

Oct. 26, 2011: EU CapitalExercise 2011 announced

Nov. 25, 2011: CDSspread widest

Dec. 8, 2011: EU Capital Exercise2011 results announced

Dec. 21, 2011: LTROintroduced by ECB

Nov. 22, 2010: EU and IMFbailout requested by Ireland

April 20, 2007: Financialsstock index peak; index = 100

March 9, 2009: Financialsstock index trough

Oct. 1, 2009: CEBS 2009results announced

July 23, 2010: CEBS 2010results announced

July 15, 2011: EBA 2011results announced

Aug. 2, 2012: OMTannounced

Apr. 23, 2010: EU and IMFbailout requested by Greece


Credibility and C


Sources: Bloomberg; European Banking Authority (EBA); Central Bank of Ireland; various financial media; and authors’ calculations.Note: CDS = credit default swap; CEBS = Committee of European Banking Supervisors; EU = European Union; LHS = left- hand side; PCAR = Prudential Capital Assessment Review; RHS = right- hand side.

Figure 14.4 Ireland: The Pain before the PCAR (Indexed to 100 on February 21, 2007)

Stock return volatility(RHS)

Event Irish Stock ExchangeFinancial Index (LHS)

Ireland CDS USD seniorfive-year (inverted, LHS)


PCAR 2011 resultsstock index level

0

120

20

80

40

60

100

0

12

6

2

10

4

8


Nov. 22, 2010: EU and IMFbailout requested

Dec. 23, 2010: AlliedIrish Banks nationalized

Jan. 7, 2011: PCAR 2011exercise announced

Mar. 31, 2011: PCAR 2011results announced

Sep. 22, 2011: Financialsstock index trough

Sep. 23, 2011: CDS spreadwidest

Feb. 21, 2007: Financialsstock index peak; index = 100

Jan. 15, 2009: Anglo Irish Bank nationalized


July 23, 2010: CEBS 2010 results announced



Li Lian Ong and C


Sources: Banco de España; Bloomberg; European Banking Authority (EBA); various financial media; and authors’ calculations.Note: CDS = credit default swap; CEBS = Committee of European Banking Supervisors; EFSF = European Financial Stability Facility; FSAP = Financial Sector Assessment Program; LHS = left- hand side; MoU = memo of understanding; OW = Oliver Wyman; RHS = right- hand side.

Figure 14.5 Spain: The “Floor” under the FSAP (Indexed to 100 on February 14, 2007)

Stock returnvolatility (RHS)

Event MSCI Spain FinancialsIndex (LHS)

Spain CDS USD seniorfive-year (inverted, LHS)


FSAP stockindex level

OW results stockindex level

0

120

20

80

40

60

100

0

8

1

5

2

3

7

4

6


May 21, 2012: Top-down stresstest announced

Jun. 8, 2012: IMF’s SpainFSAP report published

Jun. 21, 2012: Top-down stresstest results announced

July 24, 2012: Financials stock indextrough and CDS spread widest;

MoU with EFSF signed,to include bottom-up stress test

Feb. 14, 2007: Financialsstock index peak; index = 100


July 23, 2010: CEBS 2010results announced


Sep. 28, 2012: Top-down stresstest results announced


Credibility and C


TABLE 14.4

Crisis (and Follow-Up) Stress Tests: Effectiveness Scorecard

Indicator

Effectiveness of Stress Test

United States European Union Ireland Spain

Crisis Supervisory Crisis Crisis Surveillance Crisis

Instrument Measure Desired Change SCAP 2009

CCAR 2011

CCAR 2012

CCAR & DFA 2013

CEBS 2009

CEBS 2010

EBA 2011

PCAR 2011 + IMF

FSAP 20121

TD 2012

BU 2012

Financials stock index

Index return Approximately stable or rises

Return volatility Falls Credit default

swapSpread Approximately

stable or narrows 2 2 2

Effectiveness

Source: Authors.Note: BU = bottom up; CCAR = Comprehensive Capital Analysis and Review; CEBS = Committee of European Banking Supervisors; DFA = Dodd-Frank Wall Street Reform and Consumer Protection Act; EBA = European Banking Authority; FSAP = Financial Sector Assessment Program; SCAP = Supervisory Capital Assessment Program; TD = top down.1Included for completeness only—not intended as a crisis stress test; surveillance stress testing exercise was conducted in a crisis environment.2Driven by US fiscal deficit and debt-ceiling concerns.



That said, the decision as to what constitutes an “opti-mal” moment for introducing a crisis stress test is not clear- cut and remains largely an issue of judgment and, possibly, serendipity. As an extension of the principle espoused in IMF 2012a that market views should be taken into account in designing stress tests, indicators such as stock prices, their corresponding price- to- book (PB) ratios, and sovereign CDS spreads could potentially be used as triggers in deciding on the timing of a crisis stress test (Table 14.6). However, the evidence to date is inconclusive:

• The United States was first off the rank with the SCAP following two years of decline from the Febru-ary 2007 historical peak of the S&P 500 Financials Sector Index. The EU CEBS 2009 stress test was also introduced almost two years after the STOXX Eu-rope 600 Banks Price Index peaked but was less effec-tive by comparison. The Ireland and Spain crisis stress tests took place four and five- and- a- half years after the apex of their respective financials stock prices. Assuming that the decisions to stress test were made around the end of the year prior to each crisis stress test, the US and European indices would have dropped by anywhere between 65–75 percent by that stage. The long- term ( five-, 10-, and 15-year) average index levels also do not provide any clear guide to the decision- making process by the authorities, as they do not appear to have been used as trigger points. Ire-land and Spain conducted their stress tests following their engagement with the Troika for financial sup-port. By that stage, Ireland’s banks had lost almost all their market value, while the equity values of Spanish banks were down by more than 60 percent.

• The PB ratio, which is typically used to assess bank valuations, may yield some hints on the timing of the crisis stress tests. These ratios were richest in the late- 1990 s- to- early- 2000s period for the sample jurisdic-tions, reaching 3.5 times for the US financials and exceeding four times in Ireland and Spain. Long- term averages ranged from 1.8–2.2. The decision to conduct the SCAP would have been made when the PB ratio fell to unity, which could perhaps be considered a “line in the sand” for future reference. The other jurisdictions waited until their respective PB ratios had declined to well below unity, while Ireland’s PCAR would have been contemplated about the time when banks’ average PB ratio had dropped to below 0.3 times.

• Sovereign CDS spreads are an indicator of the mar-ket’s current perception of sovereign risk. Given the systemic importance of the banking sector for eco-nomic activity, market concerns that the government may have to bail out institutions that are too big to fail— and the resulting burden on the fiscal balance— are likely to be reflected in the CDS spreads. In Europe, the sovereign- bank feedback loop from banks’ large holdings of sovereign debt in-creased the likelihood of losses. Here, any rule of

include AQRs, separate liquidity stress tests, and/or follow- up solvency stress tests, some or all of which may be crucial for the credibility of the original exercise itself. The study’s overall findings are summarized in Table 14.5 in a design “scorecard” comparing the features of various elements across crisis stress tests, with the associated details presented in Appendix 14.1.

Key Elements

Timing

The timing of a crisis stress test is crucial. Steps to reduce uncertainty through information provision should be taken as soon as possible during a crisis. Borio, Vale, and von Peter 2010 posit that early recognition and intervention would avoid hidden deterioration in conditions that could magnify the costs of the eventual resolution. Pritsker 2010 argues that while central bank actions, such as broadening the range of acceptable collateral, loan guarantees, and government- sponsored capital injections, may boost bank lending during a crisis, these actions also increase the central bank’s expo-sure to credit and market risks. Such efforts would be less costly and more effective under conditions of less uncer-tainty, that is, it would be easier to convince potential lend-ers of a bank’s solvency if better information about the scope of the problem were available early on.

Experience confirms that delay by country authorities in taking resolute action in a timely manner has eventually re-quired the incurrence of significant additional costs. First, there is the destruction of the banks’ asset values, which could take a long time to recover, if at all. Second, the reputational risk to supervisory authorities also grows when a crisis is al-lowed to fester and deepen. Third, any loss in market, deposi-tor, and creditor confidence could potentially place a significant burden on the fiscal purse, and consequently, the creditworthi-ness of the sovereign if government support becomes necessary. Combined, these factors could give rise to greater demands when the authorities finally decide to take action, notably:

• Damage to the credibility of the authorities may be too deep- seated to overcome following a lengthy crisis. A consequence could be that they may have to contract third- party stress testers and seek independent over-seers to enhance the credibility of the exercise.

• Heightened uncertainty about banks’ asset quality and concerns over increasing lender forbearance could mean a more complex, resource- intensive, and pro-tracted exercise. The stress test may have to cover a much broader sample of banks than would otherwise be necessary and possibly require additional steps, such as an AQR comprising audits and third- party expert valuations of banks’ portfolios and a DIV exercise.

• Markets are likely to impose higher standards on in-stitutions that are already under extreme pressure if they have lost all trust in the quality of assets (for example, through expectations of higher loan loss projections and larger capital buffers).


Credibility and C


TABLE 14.5

Crisis Stress Tests: Design ScorecardFramework Application to Stress Test

DesignUnited States European Union Ireland Spain

Component Element Feature Importance for Success

SCAP 2009

CEBS 2009

CEBS 2010

EBA 2011

PCAR 2011

FSAP 20121

TD 2012

BU 2012

Effectiveness . . . Financials stock index stabilizes/improves and returns volatility falls

Not applicable

2

Timing of exercise

. . . Stress test is conducted sufficiently early to arrest the decline in confidence

3 Not applicable

Governance Oversight Oversight is provided by a third party

Not applicable

Stress tester(s) Stress test is conducted by third party

4

Scope Approach Stress test approach is bottom up

5

Coverage Stress test covers at least the systemically important banks and the majority of banking system assets

6 7

Scenario design

Scenarios Stress test applies large scenario shocks (2 standard deviations or larger)

Risk factors Stress test applies shocks to key risk factors

8 8 8 9 9 9

Assumptions Scenarios are standardized across banks

Behavioral assumptions are standardized across banks

Capital standards

Hurdle rate(s) Stress test applies very high hurdle rate(s) (CT1 > 6 percent)

(continued)


Li Lian Ong and C



Crisis Stress Tests: Design ScorecardFramework Application to Stress Test

Component Element Design United States

European Union Ireland Spain

Feature Importance for Success

SCAP 2009

CEBS 2009

CEBS 2010

EBA 2011

PCAR 2011

FSAP 20121

TD 2012

BU 2012

Tran

spar

ency

Objective and action plan

Objective Stress test is associated with a clear and resolute objective

Follow-up action(s) Stress test is associated with clear follow-up action(s) by manage-ment/authorities to address findings as necessary

Financing backstop Stress test is provided with an explicit financial backstop to support the necessary follow-up action(s)

10

Disclosure of technical details

Design, methodol-ogy, and implementation

Stress test discloses information

Model(s) Stress test discloses information

11 12

Details of assumptions


Bank-by-bank results


Asset quality review

. . . Asset quality review is undertaken as input into stress test

13

(continued)


Credibility and C


Framework Application to Stress Test

Component Element Design United States European Union Ireland Spain

Feature Importance for Success

SCAP 2009

CEBS 2009

CEBS 2010

EBA 2011

PCAR 2011

FSAP 20121

TD 2012

BU 2012

Follow-up stress tests

. . . Stress test assumptions on factors that management control are standardized across banks

14 15 15 Not applicable

15

Liquidity stress test

. . . Liquidity stress test accompanies solvency stress test

16

Sources: Tables 14.3 and 14.4; Appendix 14.1; and authors. Note: AQR = asset quality review; BU = bottom up; CEBS = Committee of European Banking Supervisors; CT1 = Core Tier 1; EBA = European Banking Authoriy; FSAP = Financial Sector Assessment Program; PCAR = Prudential Capital Assessment Review; SCAP = Supervisory Capital Assessment Program; TD = top down.1Included for completeness only—not intended as a crisis stress test.2Medium-term sustainability of market confidence remains to be seen. 3Delay may impose significant additional costs in order to be effective. 4Forecast losses provided by third party.5Not necessary if top down is conducted on individual banks.6Delay may result in wider coverage of banks to allay increased doubts. 7Large cross-border banks; domestic systemically important banks making up at least 60 percent of national banking assets. 8Stress test did not include haircuts to sovereign debt holdings in the banking book. 9Takes into account the European Central Bank’s Long-Term Refinancing Operation support facility.10Crisis program with the Troika.11Not critical only if independent cross-checks/validation conducted. 12Not critical only if independent cross-checks/validation conducted. 13Lower-intensity, quantitative substitute for AQR. 14Assumptions must be sufficiently stringent and must be disclosed.15Undertaken in the context of the Single Supervisory Mechanism’s 2014 Comprehensive Assessment and the EBA 2014 EU-wide stress.16The EBA conducted a confidential thematic review of liquidity funding risks.



Li Lian Ong and C


TABLE 14.6

Crisis Stress Test Jurisdictions: Financial Markets StatisticsIndicator Statistic

Instrument Measure Timing United States Europe Ireland SpainFinancials stock

indexLevel Historical peak 509.6

(Feb. 20, 2007)538.8 (Apr. 20, 2007)

17,951.5 (Feb. 21, 2007)

158.8 (Feb. 14, 2007)

End of year prior to first crisis stress test

168.8 (Dec. 31, 2008)

151.1 (Dec. 31, 2008)

414.3 (Dec. 31, 2010)

62.1 (Dec. 31, 2011)

Change from peak (percent) -67.9 -72.0 -97.7 -60.9Average level to end of

year prior to stress test5-year 398.7

(2004–2008)393.5 (2004–2008)

7,528.5 (2006–2010)

101.8 (2007–2011)

10-year 368.5 (1999–2008)

365.5 (1999–2008)

8,427.3 (2001–2010)

100.4 (2002–2011)

15-year 309.0 (1994–2008)

306.5 (1994–2008)

7,434.5 (1996–2011)

91.7 (1995–2011)

Price-to-book ratio of financials stock index

Ratio Historical peak 3.50 (Sep. 12, 2000)

2.21 (May 15, 2002)

4.16 (Jan. 1, 1999)

4.74 (Jul. 17, 1998)


0.99 (Dec. 31, 2008)

0.73 (Dec. 31, 2008)

0.26 (Dec. 31, 2010)

0.72 (Dec. 31, 2011)

Change (percent) -71.7 -67.0 -93.8 -84.8Average ratio to end of

year prior to stress test5-year 1.84

(2004–2008)1.75 (2004–2008)

1.25 (2006–2010)

1.36 (2007–2011)

10-year 2.24 (1999–2008)

. . . 1.83 (2001–2010)

1.71 (2002–2011)

15-year 2.21 (1994–2008)

. . . . . . . . .

Credit default swap Spread (basis points) Historical narrowest (based on data availability)

5.8 (Apr. 29, 2008)

46.0 (Sep. 29, 2009)

16.1 (Mar. 24, 2008)

1.5 (Jun. 20, 2005)


67.4 (Dec. 31, 2008)

. . .(Dec. 31, 2008)

608.7 (Dec. 31, 2010)

380.4 (Dec. 31, 2011)

Change from narrowest +61.6 . . . +592.6 +378.9Change from narrowest

(percent)+1,062.1 . . . +3,780.7 +25,260.0

Sources: Bloomberg; and authors’ calculations.



to adequately backstop and address the findings. The evidence suggests that while the timing of crisis stress tests may be im-portant, it is insufficient in the absence of other key elements, as elaborated throughout the rest of this section.

Governance

There is no hard- and- fast rule as to who should oversee and/or conduct the crisis stress test. The overriding requirement is that the protagonists are considered credible. In some cases, issues such as expertise, sufficiency of resources, and/or political economy considerations play equally important roles in determining who they should be.

In the United States, the oversight and execution of the SCAP relied on collaboration across supervisory agencies— the Federal Reserve, the Federal Deposit Insurance Corpo-ration, and the Office of the Comptroller of the Currency. Supervisors of individual banks were consulted but not in-volved in the actual stress test analyses.

The EU- wide stress tests were conducted by national su-pervisory authorities, overseen and coordinated by the EBA (which did not have direct interaction with the banks prior to or during the exercise) in cooperation with the European Commission and the ECB/European Systemic Risk Board. However, the EBA had argued that it needed more legal powers over the exercise to ensure the reliability of the input data, and hence the results (see Brunsden 2012).

In contrast, the authorities in Ireland and Spain ap-pointed third- party contractors in their efforts to strengthen perceptions of independence and objectivity in the process. The reputation of the supervisors had been dented after their banks passed the CEBS/EBA stress tests only to require sig-nificant restructuring not long afterward. In the case of Spain, the authorities, the Troika, the EBA, and counter-parts from two other European central banks were involved in the oversight of the stress testing exercises.

thumb that may have been used is less clear; spreads had ballooned to unprecedented levels across the board by the time any decision would have been taken on running the tests.

Ideally, a crisis stress test should be conducted before the crisis of confidence in the banking system becomes en-trenched. However, the successful exercises reveal little in terms of whether they had been appropriately timed, given that counterfactuals are difficult to prove.

By all measures, the “intervention” by the US authorities did halt and turn around the sharp slide in market confidence. That said, the rebound from the 80 percent loss in banks’ mar-ket value had been sluggish compared to the overall market, which recovered all of its losses from the crisis (Figure 14.6). The question then is whether the rise in the financials index would have been quicker and stronger had the supervisors stepped in earlier. Although bank stocks may have arguably been overvalued prior to the crisis, their overall PB ratio re-mained well below the 15-year average during this period.

The eventual outcomes from the Ireland and Spain stress tests have also been positive, but these achievements were al-most pyrrhic. The supervisors were perceived to have lost sig-nificant credibility with markets by that stage (for example, see Irish Times 2010; Garicano 2012). External consultants had to be employed to conduct comprehensive AQRs, and in the case of Spain, to run the stress tests in order to reassure investors ( third- party consultants provided forecast losses for the Ireland stress test). Moreover, the fiscal cost of supporting their respec-tive banking systems had become so onerous that both coun-tries had to eventually request external financial aid.

Irrespective of the timing of a crisis stress test, recognition of the problem alone is insufficient. It should be linked to re-structuring if a bank’s profitability is to eventually be restored. In other words, the decision to conduct a crisis stress test should also take into account the potential implications for the public purse, that is, it must be tied to the capacity of the authorities

Sources: Bloomberg; and authors’ calculations.

Figure 14.6 United States: S&P 500 Stock Market, and Financial Sector Indices (Indexed to 100 on February 20, 2007)

0

120

20

80

40

60

100


S&P 500 Index

S&P 500 FinancialSector Index



from a comprehensive AQR exercise. Specifically, the test drew on information derived from exter-nal reviews by independent auditors and real es-tate appraisers and from BdE central databases, to estimate individual banks’ capital needs under a baseline scenario and an adverse scenario (Oliver Wyman 2012a). Structural analysis of individual banks’ financial statements and business plans was undertaken. Given that the banks only ran their own models on the baseline scenario to generate net revenues, it was essentially another TD exercise— albeit at a much more granular level— but is widely referred to as a BU exercise. (For dif-ferentiation purposes, this chapter refers to the first as the TD test and the second as the BU test.)

The coverage of banks should capture at least the systemi-cally important institutions, given the macroprudential na-ture of the stress test (see IMF 2012a; Jobst, Ong, and Schmieder 2013). In this respect, guidance has been pro-vided by the Financial Stability Board on what constitutes global and domestic systemically important banks (BCBS 2011, 2012). However, the sample may have to be expanded depending on the environment at the time of implementa-tion. While some banks are of obvious systemic importance and their selection is indisputable, the difficulty has been in identifying those that are systemic at the margins, for ex-ample, some of the smaller institutions that may have the potential to become or have become systemic under certain conditions (IMF, Financial Stability Board, and Bank for International Settlements 2009). In cases where there has been a total loss of confidence in the entire banking system and uncertainty about the soundness of individual banks is very high, the coverage may have to include even the smaller, nonsystemically important banks to forestall a “witch hunt” for failed and failing banks. Coverage differed across the various crisis stress tests (including whether the tests were run on consolidated or domestic business data), but each ex-ercise captured at least 60 percent of domestic banking sys-tem assets:

• The US SCAP included the 19 largest bank holding companies, each with total assets greater than $100 billion. They represented two thirds of banking sys-tem assets.

• The EU CEBS 2009 stress test captured 22 large cross- border banks with 60 percent of EU banking assets; the number of banks increased to 91 in subse-quent exercises, covering 21 countries and at least 50 percent of each banking sector, for an additional 5 percentage point coverage of EU banking assets. However, the flexibility for country authorities to choose which banks to include in the stress tests was perceived to have reduced the legitimacy of the exercises (Ahmed and others 2011).

• Ireland’s PCAR 2011 stress tested four financial in-stitutions, which accounted for 80 percent of bank-ing system assets.

Scope

There is some flexibility to the stress testing approach taken in a crisis exercise. Ideally, a bottom- up (BU) test, cross- validated by a top- down (TD) exercise, would be the superior approach (IMF 2012a; Jobst, Ong, and Schmieder 2013), but this may not be possible in a crisis situation where the timeframe is compressed (see Figure 14.1, footnote 1 for the IMF staff’s definitions of BU and TD stress tests). Both BU and TD ap-proaches have been used effectively in crisis stress tests. How-ever, if only a TD stress test can be undertaken, it should be conducted on a bank- by- bank rather than an aggregated basis, which is consistent with the need for transparency at the dis-closure stage, as is discussed later in this section. Additionally, the stress tests should be supported by inputs from AQRs (and preferably DIVs), which is covered later in this chapter.

• The US SCAP consisted of a BU and TD mix, with what would be deemed a lower- intensity, quantitative substitute for an AQR. The supervisors applied indepen-dent quantitative methods using firm- specific data to support their assessments of banks’ submissions (Board of Governors of the Federal Reserve System 2009a).

• The EU CEBS 2010 and the EBA 2011 exercises com-prised BU tests by cross- border banking groups and simplified stress tests, based on national supervisors and reference parameters provided by the ECB, for less complex institutions. The CEBS 2010 stress test included a peer review of the results and a challenging process, as well as extensive cross- checks by the CEBS (CEBS 2010b); the process evolved for the EBA 2011 exercise to incorporate consistency checks by the EBA, a multilateral review and TD analysis by the EBA and the European Systemic Risk Board with ECB assis-tance (EBA 2011c).

• Ireland’s PCAR 2011 was a BU exercise supported by an AQR. Banks were required to model the im-pact of certain assumptions on their balance sheets and profit- and- loss accounts (revenues and losses) based on a third party’s assessment of forecast losses (CBI 2011). The stress test was perceived to be par-ticularly credible in that it explicitly compared the loan loss estimates of the CBI with those of the third party as a cross- check of the results, which were sub-sequently published by the CBI.

• In Spain, two sets of crisis stress tests were conducted in 2012 and the results were published by the third- party consultants who conducted the exercises:– The first exercise was a TD stress test. Two con-

sultants separately considered the historical per-formance and asset mix for each institution at aggregate levels to generate forward- looking pro-jections. The consultants applied their own mod-els, expert experiences, and benchmarks (Roland Berger 2012; Oliver Wyman 2012b).

– The second stress test was conducted by one con-sultant, using detailed data from banks and inputs



1.3, and 1.5 standard deviations from their respective baseline growth scenarios (Box 14.1), with attendant shocks to other macroeconomic variables. However, the test results did not gain wide acceptance.

• What is not commonly known is that the adverse growth scenario used in the SCAP was even less stressful than any of the CEBS/EBA shocks. It was equivalent to a cumulative one standard deviation from the baseline over the two- year risk horizon, de-termined well before the contraction had bottomed out (Figure 14.7). Indeed, the SCAP stress scenario was criticized by some at the time the results were announced for likely being closer to the actual base-line itself (for example, Fox 2009). Yet, the SCAP was effective in regaining market confidence.

• Similarly, the growth shocks applied to both the Spain TD and BU stress tests were equivalent to one standard deviation from the projected baseline, while Ireland’s PCAR 2011 used the EBA scenarios.

The selection of macroeconomic parameters in the scenario design does not appear to significantly influence the credibil-ity of a crisis stress test either. The SCAP was parsimonious, with three variables (real GDP growth, the unemployment rate, and house prices), while the Ireland and Spain stress tests employed more than a dozen different ones (Table 14.7). Un-like the growth scenarios, and outside of some coverage of the real estate and employment variables, the projections for most of the other variables were generally less scrutinized.

Consistent with best practice, comprehensive coverage of material risk factors in crisis stress tests appears to be much more relevant for the reliability of the results (Basel Commit-tee for Banking Supervision 2008; Board of Governors of the Federal Reserve System, Federal Deposit Insurance Corpora-tion, and Office of the Comptroller of the Currency 2012; IMF 2012a). The global financial crisis has brought to the fore risks that had previously been in the periphery or had not been considered, such as exposures to sovereign and other previously low- default assets, their accounting in the banking or trading book, funding costs, and cross- border exposures, among others (Jobst, Ong, and Schmieder 2013). The US stress test covered banks’ entire balance sheets (including their international exposures) while the European stress tests focused on banks’ domestic loan books (Table 14.8), which were the main concern of investors. However, the exclusion of some important risk factors affected the credibility of some of these exercises:

• The EU stress tests have been vociferously criticized for their inadequate capture of important risk fac-tors, owing in part to political economy constraints (see Wilson 2011; Wishart 2011). The failure to properly stress banks’ sovereign exposures was con-sidered particularly egregious in light of the debt cri-sis and concerns about the bank- sovereign feedback loop (for example, Ahmed and others 2011; Das 2011; Steinhauser 2011). Specifically, the haircuts imposed on banks’ sovereign portfolios in the trad-ing book during the EBA 2011 exercise were seen to

• In Spain, the TD and BU stress tests covered banks ac-counting for around 90 percent of total system assets. Initial concerns had been with some medium- sized and smaller banks rather than the largest, most systemically important banks. However, the slow deterioration in sentiment over a prolonged period and constant revela-tions of new problems eventually affected perceptions of the entire banking system. In the end, the inclusion of both the largest banks and the smaller problem ones in both stress tests became necessary in order to differ-entiate the strong institutions from the weak ones.

Scenario Design

The selection of adverse macroeconomic scenarios in crisis stress tests represents a delicate balance between the need to be credible yet constructive. As a principle, stress scenarios should capture extreme but plausible shocks, that is, the tail risks to the financial system (BCBS 2008; IMF 2012a). However, while this principle should always be applied in stress tests for surveillance purposes to support discussions on supervisory actions and crisis preparedness (Jobst, Ong, and Schmieder 2013) and in regular supervisory stress tests, it needs to be more nuanced in a crisis situation.

In a crisis stress test, the adverse scenario should reflect the uncertainty around the baseline. Crises are typically already tail- risk events in themselves. In some cases, they may even be labeled “black swan” events at the outset, as some have argued was the case during the global financial crisis (for example, Helmore 2008; Skidmore 2008), although the prolonged ac-cumulation of economic and financial imbalances may be ob-vious in hindsight. In such an environment, banks may already be under severe stress. In other words, the point of the cycle at which the shock is applied matters. Consequently, any implementation of further “tail of the tail” shock scenarios that would hypothetically obliterate an entire banking system would obviate any constructive planning of needed follow- up action(s) by the authorities and the banks themselves. Borio, Drehmann, and Tsatsaronis 2012 argue that it is easier to identify relevant scenarios for stress testing purposes after a crisis has erupted as the system “does not need to be shaken so hard to reveal weaknesses.” Rather, a key consideration in the scenario design at that stage is that the crisis stress test should be able to differentiate across institutions, as a first step to ward determining whether a capital injection, or some other form of balance sheet restructuring or resolution, is required.

The evidence from the crisis case studies suggests that the magnitudes of the macroeconomic shock scenarios per se are not an overriding element for success. The CEBS/EBA stress tests have been derided for the apparent mildness of their ad-verse growth stress scenarios, among other things, contribut-ing in part to their lack of acceptance (for example, Ahmed and others 2011; Campbell 2011; Jenkins 2011; Steinhauser 2011; IMF 2013b). However, a closer examination of the other crisis stress tests suggests that this argument may be flawed:

• The CEBS 2009 and 2010 and the EBA 2011 exer-cises applied cumulative growth shocks averaging 1.9,



addressed in the BU exercise. Auditors and real es-tate appraisers were appointed to verify the quality of the input data. The issue of sovereign risk was omit-ted but was considered less of an issue owing to the availability of the LTRO facility from the ECB by the time of the stress tests. The liquidity support al-layed market concerns about banks’ funding costs and possible deep haircuts to their sovereign debt

have been too lenient, as they only applied a market value adjustment rather than possible defaults, while the omission of any stress test of the banking book— where the majority of banks’ sovereign exposures resided— meant that the main risk factor at the time had not been adequately captured.

• In Spain, concerns about lender forbearance and possible misclassifications in banks’ loan books were

Box 14.1. Designing Crisis Stress Test Growth Scenarios

The Committee of European Banking Supervisors stress tests popularized the notion of calibrating growth shocks in terms of the number of standard deviations from the baseline. This metric allows for a more standardized comparison across stress tests at a point in time and over time. For instance, the application of a large growth shock scenario to an economy that typically experiences large and volatile growth rates may be a less significant event than to one that consistently posts more moderate and stable growth. The Committee of European Banking Supervisors method for determining adverse growth scenarios consists of the following steps:

1. Calculate the two- year growth rates over the preceding 30 years.2. Calculate the standard deviation of the two- year growth rates over the 30-year period.3. Calculate the desired number of standard deviations of the two- year growth rate.4. Apportion the standard deviation(s) growth over the two- year horizon and subtract from each year of the baseline forecast to derive

the stressed scenario.The rule of thumb in the IMF’s Financial Sector Assessment Program scenario stress tests has generally been to apply two standard de-

viation shocks to growth (IMF 2012a; Jobst, Ong, and Schmieder 2013), but calibrations have been necessary in crisis situations. For exam-ple, Spain’s Financial Sector Assessment Program stress test imposed a “severe adverse” scenario of one standard deviation from the baseline GDP growth trend over a two- year risk horizon (IMF 2012b). The shock came on top of a downward adjustment to the World Eco-nomic Outlook baseline forecast to incorporate downside risks to growth from the crisis plus a projected fiscal adjustment. In this scenario, most of the shock to the baseline growth (about two thirds) was assumed to occur in the first year, attributable to a sharp decline in output, further declines in house prices, and rising unemployment.

Viewed from another angle, the two- year cumulative GDP shock for Spain under the severe adverse scenario was considered extreme by historical standards, as the actual outcome proved. The GDP drop in the first year of the risk horizon approximated the largest decline in economic activity since 1945 but represented a plausible “tail of the tail” risk under the circumstances (Figure 14.1.1). The third- party crisis stress tests subsequently increased the second- year stress and extended both scenarios to a third year. As it turned out, the actual growth in 2012—the first year of the risk horizon— approximated the projected baseline.

A corroborating method to gauge the extremity of a proposed shock scenario is to determine its deviation from the long- term historical av-erage, in standard deviation terms. In the Spain example, the adverse shock scenario extended beyond three standard deviations of the mean annual growth rate of the past 30 years; it exceeded even the sharp contraction experienced in 2009 and was designed to be more protracted.

Sources: Oliver Wyman 2012b; World Economic Outlook; and authors’ estimates.Note: std. devns. = standard deviations.

Figure 14.1.1 Spain: 30-Year Average Annual Growth Rate and Stress Test Scenarios (In percent)

Actual Baseline IMF adverse Third-party adverse30-year meanannual growth rate

1 std. devnfrom mean

2 std. devns.from mean

3 std. devns.from mean

–5

6

–4–3–2–1

0

4

2

5

3

1

1981 0905200197938985 13



Sources: Federal Reserve; and authors’ estimates of annualized quarterly growth profiles for both SCAP 2009 scenarios and the CCAR 2011 baseline scenario.Note: CCAR = Comprehensive Capital Analysis and Review; DFA = Dodd- Frank Wall Street Reform and Consumer Protection Act; SCAP = Supervisory Capital Assessment Program.

Figure 14.7 United States: Baseline and Adverse Growth Scenarios for Crisis and Supervisory Stress Tests (In percent, quarter- over- quarter annualized)

SCAP 2009 Baseline SCAP 2009 Adverse CCAR 2011 BaselineCCAR 2011 Adverse CCAR 2012 Baseline CCAR 2012 Adverse

CCAR & DFA 2013 Baseline DFA 2013 AdverseCCAR & DFA 2013 Severe Adverse Actual

–10

6

–8

–6

–4

–2

0

4

2

2006:Q1 14:Q113:Q112:Q111:Q110:Q109:Q108:Q107:Q1 15:Q1

TABLE 14.7

Crisis Stress Tests: Macro-Financial Parameters ScorecardParameter Application to Stress Test

Variable Indicator United States European Union Ireland Spain

SCAP 2009

CEBS 2009

CEBS 2010

EBA 2011

PCAR 2011

FSAP 20121

TD 2012

BU 2012

Growth Real GDP Real GNP Nominal GDP

Employment Unemployment rate Employment

Price evolution CPI 2 HICP GDP deflator

Consumption Private Government

Trade Exports Imports Balance of payments

Income and investment

Investment Personal disposable income

Real estate Real estate prices Commercial property Residential property Land

Interest rates Short-term interest rate (12 months or less)

2

Medium-term interest rate (up to five years)

2

Long-term interest rate (more than five years)

2

Exchange rate Relative to US dollar 2

Stock market Stock price index 2 Credit to other

resident sectorsHouseholds Nonfinancial corporate

Sources: Central Bank of Ireland; European Banking Authority (EBA); Federal Reserve; IMF; Oliver Wyman; and Roland Berger.Note: Even though some variables (for example, commodities, credit default swaps, securitized assets) were not provided as part of the general macrosce-narios, they were used in the determination of key market risk drivers. BU = bottom up; CPI = consumer price index; CEBS = Committee of European Bank-ing Supervisors; EBA = European Banking Authority; FSAP = Financial Sector Assessment Program; GNP = gross national product; HICP = Harmonized Index of Consumer Prices; PCAR = Prudential Capital Assessment Review; SCAP = Supervisory Capital Assessment Program; TD = top down.1Included for completeness only—not intended as a crisis stress test; surveillance stress testing exercise was conducted in a crisis environment.2Information not disclosed.



TABLE 14.8

Crisis Stress Tests: Risk Factors ScorecardRisk Factor Application to Stress Test

Risk Type Nature of Accounting

Exposure United States

European Union Ireland Spain

SCAP 2009

CEBS 2009

CEBS 2010

EBA 2011

PCAR 2011

FSAP 20121

TD 2012

BU 2012

Credit risk . . . Residential mortgages 4 First lien

Second lien Commercial and

industrial loans2

Corporate loans 3 4 RE developers SME loans 3 CRE loans Financial institution

loans 4

Consumer loans (including credit card)

4

Revolving loans 3 Public works 3 Sovereign exposure in

available-for-sale (AfS) banking book

Other loans Market risk Trading book Sovereign portfolio 4

Financial institutions portfolio

4

Other securities (incl. MBS and other ABS)

Private equity holdings Counterparty credit

exposures to OTC derivatives

Banking book (AfS)

Sovereign portfolio



Banking book (HtM)

Sovereign portfolio



Operational risk

. . . . . .

Separate liquidity risk test

. . . . . . 5

Sources: Central Bank of Ireland; European Banking Authority (EBA); Federal Reserve; IMF; Oliver Wyman; and Roland Berger.Note: ABS = asset-backed securities; BU = bottom up; CEBS = Committee of European Banking Supervisors; CRE = commercial real estate; FSAP = Financial Sector Assessment Program ; HtM = hold to maturity; MBS = mortgage- backed securities; OTC = over the counter ; PCAR = Prudential Capital Assessment Review; RE = real estate; SCAP = Supervisory Capital Assessment Program; SME = small- and medium-sized enterprise; TD = top down.1Included for completeness only—not intended as a crisis stress test; surveillance stress testing exercise was conducted in a crisis environment.2Includes corporate, SME, revolving, and public works loans.3Included under “commercial and industrial loans.”4Information not disclosed.5The EBA conducted a confidential thematic review of liquidity funding risks.



of stress, the scenario- based stress testing framework incorpo-rates quantitative forecasts for a wide range of macroeconomic variables to generate a wide range of plausible outcomes. It fa-cilitates the identification of risk concentrations in the bank-ing system and, consequently, preparedness for dealing with the dangers of an uncertain future. Langley 2013 argues that the “practical usefulness” of the SCAP’s results, which made it possible for the authorities and the banks to act on an antici-pated financial future— rather than the actual results them-selves (which in fact showed that the major banks needed to raise significant additional capital) or their accuracy (which cannot be proven ex ante)—underpinned its success.

In a similar context, crisis stress testing has also placed the spotlight on the modeling of revenues and losses. While stress testing for losses has typically been to map macrofactors onto the various risk factors that drive the impairment parameters, the crisis has underscored the importance of adequately mod-eling losses for different categories of credit risk (for example, various types of real estate, corporate sector, and credit card loans), geographic heterogeneity, and a rapidly evolving macro- financial environment for which there has been no precedent. Separately, stress testing revenues— especially for stressed conditions— is largely seen to be a “black box” (Schuermann 2012). Given the importance of projected pre-provision profits in determining banks’ loss- absorption ca-pacity in stress scenarios, the credibility of these estimates is key in the overall perception of any stress testing exercise.

Capital Standards

The capital standards applied to crisis stress tests play a cru-cial role in their legitimacy, but evidence from the case stud-ies suggests that some variability is acceptable. Countries would typically apply their existing capital frameworks. In this context, the differences in regulatory frameworks and thus difficulty in comparing stress test results across jurisdic-tions do not appear to be an overriding concern for markets, as long as the definition of capital is made clear in each case. Bernanke 2010 notes the importance of focusing not just on the levels of capital but also on the composition of capital (which is also consistent with Basel III) in a crisis stress test:

• The US authorities applied their existing capital frame-work. Banks were required to meet the T1 capital hur-dle of 6 percent post stress and the higher quality T1 common equity ratio of 4 percent post stress. Basel I risk weights were used to calculate risk- weighted assets (RWA), providing transparency in this aspect of the stress test. Nonetheless, the authorities acknowledged in designing the SCAP that “no single measure of capi-tal adequacy is universally accepted or would guarantee a return of market confidence” (Bernanke 2009).

• The EU, Ireland, and Spain stress tests applied the existing Capital Requirement Directive at the time (that is, Capital Requirement Directive II) to the cal-culation of capital. The Basel II risk weights– which are more opaque— were used to calculate RWA. That

portfolios as the pressure for banks to liquidate their holdings in the hold- to- maturity banking book and realize the losses abated.

Another aspect of crisis stress testing is the standardization of assumptions and not just the assumptions themselves. Crisis stress tests tend to be more constrained in the assumptions that are employed, in order to facilitate comparisons. That said, ab-solute standardization is not necessary for credibility. All crisis stress tests have imposed consistent macroscenario(s) on all banks within a particular jurisdiction, but behavioral assump-tions (that is, assumptions about factors that management con-trols) have been allowed to vary, typically with cross- checks by another party to ensure their reasonableness. Ultimately, what has been more important is the publication of information re-lating to those assumptions so that market participants are able to replicate the results to their own satisfaction (see discussion on communication later in this section):

• In the SCAP, the US supervisors provided assump-tions for the macroeconomic scenarios. Banks were asked to adapt the assumptions to reflect their spe-cific business activities when projecting their poten-tial losses and resources for absorbing those losses; supervisors then reviewed and assessed the firms’ submissions and the quantitative methods that were used to project those losses and resources, as well as the key assumptions (Board of Governors of the Fed-eral Reserve System 2009a, 2009b). To facilitate horizontal comparisons across firms, supervisors ap-plied their own independent quantitative methods to firm- specific data.

• The EU CEBS/EBA stress tests applied macroeco-nomic and sovereign shock scenarios and parameters developed by the ECB. Very detailed and prescriptive guidance on assumptions and methodologies were provided for the EBA 2011 exercise (EBA 2011c). Banks’ calculations were reviewed and challenged by the respective national supervisors, then analyzed by the EBA, which conducted in- depth consistency checks and challenges with national supervisors.

• Ireland’s PCAR 2011 incorporated many of the pa-rameters used for the EBA 2011 stress test. A private consultancy firm was contracted by the CBI to pro-vide oversight, challenge the work of the third- party stress tester, and ensure consistency across institu-tions and portfolios (CBI 2011).

• The Spanish stress tests used the growth scenarios and guidelines provided by a steering committee comprising the authorities, the Troika , and counter-parts from two European central banks. The process and methodology for the BU exercise were closely monitored and agreed upon with an expert coordi-nation committee from the Troika, the EBA, and the authorities (Oliver Wyman 2012a).

The crisis brought forward- looking techniques to the front and center of stress testing. Eschewing in part the backward- looking probabilistic calculations based on historical periods



• The former should not risk prejudging the final result— the underlying conditions of banks need to be determined first.

• The latter must be in place at the time of the com-mencement of the exercise to avoid any uncertainty on the part of depositors or investors who may be concerned about their holdings of bank debt or the possible dilution of their shareholdings. It should comprise a clear action plan, and credible financial backstops against possible adverse findings must be at hand (see Schuermann 2012). For instance, the revelation of a potentially large gap in bank capital-ization with no market access would require other ready sources of funding to fill that capital need.

• In some cases, the restoration of solvency may re-quire a detailed roadmap for significant balance sheet and cost restructuring. Merely raising capital would be ineffective if cleaning up balance sheets is necessary for their repair (see Borio, Drehmann, and Tsatsaronis 2012). Importantly, any restructuring should be carried out swiftly and, as much as possible, in ways that do not worsen sovereign debt burdens (Claessens and others 2011).

• In other cases, the resolution of nonviable banks may be necessary to ensure the future health of the sys-tem. Thus, having an adequate resolution framework in place to take the requisite action is also key to any successful outcome arising from crisis stress tests.

In these areas, the design and execution of crisis stress tests have varied across jurisdictions:

• The US authorities are generally perceived to have stood “wholeheartedly” behind their stress test results (Onado and Resti 2011). The SCAP was designed and implemented to meet a clearly defined policy ob-jective with the necessary financial support.– The authorities explicitly noted that the aim of the

SCAP was to try to change macroeconomic out-comes by ensuring that the largest banks had suffi-cient capital buffers so that they would remain well- capitalized and be able to continue providing credit and intermediation services even in an eco-nomic environment that was more challenging than anticipated at the time (Board of Governors of the Federal Reserve System 2009c; Tarullo 2010).

– At the start of the exercise, the authorities an-nounced that banks needing to augment their capital poststress test would be given one month to design a detailed plan and six months to raise the requisite extra capital, and that they would be bridged by the US Treasury’s firm commitment to provide contingent common equity under the Capital Assistance Program of the Troubled Asset Relief Program in the meantime.

– Clarifications (or “forward guidance”) by the au-thorities that the SCAP would not be used as a pretext for government takeovers of the largest

said, the capital definition deviated from that of the regulatory directive:– The CEBS 2009 and 2010 stress tests applied a

T1 hurdle rate of 6 percent. The EBA 2011 stress test evolved in line with Basel III developments— it implemented a commonly agreed- upon defini-tion of common equity capital (“EBA Core Tier 1 [CT1]”) and applied a poststress hurdle rate of 5 percent, noting that a higher threshold than the legal minimum was “necessary in assessing the resil-ience of banks in adverse circumstances if credibility and confidence in the banking sector is to be re-stored” (EBA 2011d).

– Ireland imposed an EBA CT1 hurdle rate of 10.5 percent for the baseline scenario and 6 percent under stress (up from the 4 percent required min-imum), plus an additional protective buffer.

– The Spain TD and BU stress tests applied an EBA CT1 hurdle rate of 9 percent under the baseline scenario and 6 percent for the adverse scenario.

In a crisis situation, tensions may arise between micro-prudential and macroprudential objectives in determining the adequacy of capital buffers (IMF 2013a). While con-cerns such as lender forbearance and loan misclassification should be taken into account, especially in instances where AQRs are not undertaken, requiring banks to hold very high poststress test capital ratios (microprudential) to meet— sometimes unreasonable— market expectations (see Box 14. 2) could lead to excessive deleveraging, forestall the issuance of new credit to the economy, and exacerbate the economic downturn (macroprudential). The result could be a vicious circle of further deterioration in the asset quality of banks, and consequently, further destruction of capital.

Instead, banks should build strong prudential buffers during good times so that they are in a position to reduce them during bad times in a manner that respects micropru-dential objectives. The design of the capital standards for the Ireland and Spain crisis stress tests applied this philosophy— banks were expected to maintain a high level of CT1 capital adequacy (which incorporated a buffer) under a central (baseline) case scenario, but were assumed to be able to draw on the buffer in the event that an extreme stress scenario were to materialize. During bad times, encouraging in-creases in capital levels rather than ratios could align both microprudential and macroprudential objectives.

Transparency

Objective, Action Plan, and Financial Backstop Crisis stress tests provide the financial foundation for authorities to take necessary action(s) to restore stability to the banking sector. The ultimate overarching objective should be to en-sure that the financial system returns to health and that the recovery is durable. Thus, any crisis stress test should be de-signed to meet a well- specified policy goal, accompanied by a comprehensive strategy to address the findings:



Box 14.2. The Potential Impact of Capital Hurdle Rates for Crisis Stress Tests

In a crisis stress test, where the results may require follow- up capital action, the setting of capital hurdle rates is of significant import. Com-bined with the magnitude(s) of the applied shock(s), hurdle rates play a potentially crucial role in estimating any required recapitalization, and consequently, in any decision to restructure or exit banks from the system. These could have far- reaching implications for the raising of capital and possibly the fiscal budget.

The recapitalization of banks based on stress test outcomes could affect their lending capacity if the hurdle rates are set too high. As a simple example (Figure 14.2.1), assume that a bank has (1) a preshock Core Tier 1 (CT1) ratio of 9 percent; (2) constant risk- weighted assets; and (3) to meet a required post-sress CT1 capital adequacy hurdle rate of 11 percent, which includes a buffer. Next, consider two stress test scenarios— a baseline and an adverse:

1. Baseline (central case)(a) Assume that under the baseline scenario, the bank’s CT1 ratio is reduced by 2 percentage points to 7 percent.(b) The bank is expected to take capital action that would return the CT1 ratio up to 11 percent, that is, an increase of 4 percentage

points.(c) In other words, the bank would have to “top up” its existing 9 percent CT1 ratio with another 4 percentage points up to 13 per-

cent, in anticipation of the baseline scenario materializing.(d) This means that the bank would have to hold a total capital adequacy ratio of more than 16 percent, once additional requirements

to make up Tier 1 and total capital are included, and even before taking into account possible items such as domestic systemically important banks or global systemically important bank surcharges.

(e) If the central case growth forecast is accurate and the bank’s CT1 ratio is indeed reduced by 2 percentage points, the bank would have a CT1 ratio of the targeted 11 percent.

2. Adverse(i) Assume that under a severe adverse scenario, the tail shock sharply increases the bank’s projected losses and reduces its CT1

ratio by 6 percentage points to 3 percent.(ii) The bank is expected to take capital action that would return the CT1 ratio back up to 11 percent, that is, an increase of 8 percent-

age points.(iii) In other words, the bank would essentially have to have a CT1 ratio of 17 percent (that is, the existing 9 percent plus another

8 percentage points).(iv) This means that the bank would have to hold a total capital adequacy ratio of more than 20 percent, once additional require-

ments to make up T1 and total capital are included, and even before taking into account possible items such as domestic systemically important banks or global systemically important bank surcharges.

(v) If the baseline scenario were to materialize, the bank would be carrying a CT1 ratio of 15 percent (that is, 17 percent less the 2 per-centage point impact).

Private sector stress tests of the Spanish banking sector in 2011–2012 applied similarly stringent assumptions. Their huge estimates of the recapitalization needs of the banks were presaged on projected losses of up to half, CT1 thresholds of up to 11 percent, and a capital hole of up to €120 billion (Table 14.2.1). As it turned out, the baseline growth scenario for 2012 eventually became the actual outcome (Box 14.1).

Source: Authors.Note: CT1 = Core Tier 1; T1 = Tier 1.1 In both scenarios, the absolute amount needed to “replenish” the capital may be lower if risk- weighted assets decrease in line with the loan losses.

Figure 14.2.1 Hypothetical Recapitalization Estimations1 (In percentage points)

Initial CT1T1 add-onPoststress CT1

Total capital add-onCT1 “top-up” based on stress test

0

22

24

8

12

16

20

6

10

14

18

Baseline AdverseCurrent

ExistingCT1 ratio

(a) (b) (c) (d) (e) (i) (ii) (iii) (iv) (v)

(continued)



to any of the CEBS or the EBA stress tests and could not provide any collective financial back-stop for the results.

• In Ireland, the 2011 PCAR was undertaken following the government’s request for financial support from the Troika (see Department of Finance— Government of Ireland and CBI 2010). The government had re-quested an IMF arrangement under the Extended Fund Facility for a period of 36 months in the amount of €22.5 billion, in addition to €45 billion from the European Stability Mechanism/European Financial Stability Facility including bilateral loans and own resources, in November 2010. The stress test formed part of the agreed reforms of the domes-tic banking sector under the Financial Measures Program, the banking element of this package.

• Nowhere was the difference between having a clear ob-jective and action plan in place and not having them more obvious than in the case of Spain. Markets re-mained unconvinced following the release of the results from the TD stress test in June 2012. The exercise had been undertaken to obtain an “overall figure” for the recapitalization needs of the Spanish banking system as a precursor to a more granular evaluation of individual bank portfolios as part of its request for EU assistance (Spain Ministry of Economy and Competitiveness and BdE 2012). However, it was not accompanied by any specific details on a financial backstop or follow- up ac-tion to address the problems in the banking sector. Sen-timent only firmed upon the actual signing of the memo of understanding with the Eurogroup in July, under which financial assistance to the banking sector would be provided through the European Stability Mechanism. The aim was to use the subsequent BU

banks, if nationalization was not necessary, pro-vided support for their stock prices; indeed, the stock prices of SCAP banks outperformed the non- SCAP ones during the stress test period, pos-sibly because it was unclear how the latter would traverse the financial crisis.

• The contrast between the US and EU crisis stress tests has been stark:– The clarity of the EU objectives improved only over

time. The stated aim of the CEBS 2009 stress test was vague, with the authority initially noting that the exercise was being held in the context of su-pervisors’ regular risk assessment of the financial sector (CEBS 2009). In contrast, the objectives of the CEBS 2010 and the EBA 2011 exercises were explicit— to provide policy information about the overall resilience of the EU banking system for the assessment of banks’ resilience to adverse eco-nomic developments and to inform policymakers about the ability of banks to absorb those shocks (CEBS 2010b; EBA 2011d).

– Moreover, little guidance was provided by the EU authorities collectively on possible action plans and the availability of resources to back them. Follow- up measures to the CEBS 2010 stress test were left to individual national authorities to pur-sue. The tests failed to reassure the markets, espe-cially when some banking systems subsequently came under severe pressure. The EBA 2011 exer-cise subsequently required banks showing capital shortfalls to present their plans to restore their capital positions and to implement remedial mea-sures within six months. However, the European governments could not reach any agreement prior

Box 14.2. (continued)

TABLE 14.2.1

Spain: Market Assumptions and Estimates of Bank Recapitalization NeedsStudy Expected Losses1

(In percent)CT1 Threshold

(In percent)Recapitalization Needs

(In billions of euro) 1 14 10–11 79–86 2 11–14 Current level 65 3 9 . . . 802

4 14 10 45–55 5 16 10 33–573

6 16 9 90 7 . . . 11 68 8 . . . Current level 54–973

9 51 . . . 584

10 11–19 RDL 2/2011 45–1193

Sources: Various bank and rating agency reports. Note: CT1 = Core Tier 1; RDL = Royal Decree Law.1Average of the whole loan portfolio.2Provision shortfall (taking into account RDLs).3Baseline and stress scenario.4Data refer to the real estate portfolio only.



rather than the results per se. Although bank- by- bank results were published in the CEBS 2010 and the EBA 2011 exercises, investors were skeptical about the failure to adequately stress banks’ sovereign exposures in the banking book, as discussed previously, and consequently remained unconvinced about their health. In particular, the disclosures of the EBA 2011 stress test results were richly documented and included details on the sovereign bond portfolios of individual banks as well as their capital composition. However, the EU authorities lacked the mandate to require any follow- up action in these areas and were thus unable to allay market concerns without being able to provide clarity on this part of the exercise. It was not until the EU Capital Exercise 2011 (EBA 2011e), when banks were re-quired to disclose the requisite sovereign capital buffers against their exposures and to submit their recapitalization plans to reach 9 percent CT1 capital, that market sentiment began to bottom out (Figure 14.3; IMF 2013b).

More generally, the effective crisis stress tests were the ones that published detailed information on certain aspects of the exercise (Board of Governors of the Federal Reserve System 2009b; CBI 2011; Oliver Wyman 2012a). Specifically, they disclosed: (1) the stress test design and methodology and their implementation; (2) macroeconomic, absorption capacity, and loan loss assumptions; and (3) individual bank results showing projected losses for portfolio catego-ries considered most important by markets for a particular banking system (for example, by loan type for the United States, Ireland, and Spain), capital components, and pro-jected capital shortfalls (Table 14.9). Details of the stress test models were typically not published, but markets seemed satisfied by the detailed cross- checks and reviews conducted by the authorities or third parties.

Other Important Considerations

Asset Quality Review

Reliable inputs are critical for the credibility of any crisis stress test. Thus, an AQR of banks’ portfolios should be undertaken ahead of the stress test, although the nature and extent of the AQR may differ depending on market perceptions of the reli-ability of the reported information and the conduct of the stress test. It should also ideally include a DIV exercise to ensure the completeness and accuracy of data and the veracity of related information technology and risk- monitoring systems at banks.

In a crisis environment, an AQR would be a first step to-ward a comprehensive and detailed assessment of possible stress test buffers. It would help ensure that the input data are “clean” and thus facilitate more realistic loan loss estimates. An AQR typically comprises the following two important but different types of costs:

• The actual cost of running the exercise, which may require a third-party contractor, plus possibly input from auditors and asset valuation companies.

• The cost of cleaning up the books first if significant inaccuracies in reporting (that is, incorrect loan and/or nonperforming loan classifications) and/or

stress test to identify institutions that needed to be re-structured and to require concerted reforms of the banking sector as key conditions for financial support.

Disclosure of Technical Details Transparency is an indis-pensable requirement of any crisis stress test. Peristiani, Morgan, and Savino 2010 posit that investor uncertainty about the condition of banks during a crisis stems from sev-eral sources. These include concerns as to how banks ac-count for losses and their true capital adequacy going forward; the capital standard that regulators would apply to a bank; and how the government would deal with insolvent banks, that is, whether it would nationalize the banks and wipe out the value to investors or whether it would inject capital and mitigate investors’ losses. Goldstein and Sapra 2012 argue that a key part of the supervisory disclosure on stress tests is to hold supervisors accountable for their actions ahead of time about (1) what is needed for firms to meet the test requirements; (2) what firms that do not meet the re-quirements would be expected to do; and (3) what steps su-pervisors would take with those firms.

The public nature of crisis stress tests is premised on the desire to improve transparency into the health of individual banks and that of the banking system as a whole. The sever-ity of the global financial crisis has been attributable in part to bank opacity— excessive risks taken by banks were not adequately disclosed and markets could not distinguish the healthy banks from the weak ones during the crisis (Peris-tiani, Morgan, and Savino 2010; Goldstein and Sapra 2012). Hence, detailed, quality disclosure of bank- specific informa-tion from any crisis stress test is crucial, as it will allow inves-tors and counterparties to understand the risk drivers for each institution, improve market discipline, and reduce the risk premia charged (Pritsker 2010). The actual substance of the information should enable the market to do its own as-sessment of the scenarios, assumptions, and the resulting out-comes at the bank level (Bernanke 2010; Schermann 2012). The double- edged sword is that the disclosure of stress test results, if not properly designed, may actually create panic by introducing more noise (Goldstein and Sapra 2012).

The public disclosures surrounding the SCAP are consid-ered one of the main reasons for its success. By assessing the overall needs of the US financial system and the specific needs of individual banks, the exercise provided valuable in-formation to market participants on risk concentrations (Hirtle, Schuermann, and Stiroh 2009; Langley 2013). Peristiani, Morgan, and Savino 2010 investigate the infor-mation value of the SCAP and find that the supervisors’ comprehensive assessments of each bank’s estimated losses and capital needs under the adverse scenario produced infor-mation about the banks that private sector analysis did not already know. The upfront transparency with regard to the availability of the financial backstop then provided the nec-essary reassurance against those findings.

The EU exercise further demonstrated the importance of disclosing information relevant for addressing market concerns,



market’s concerns about the quality of banks’ loan portfolios by using the following more quantitative methodology:

• Banks were instructed to estimate cash- flow losses using a set of indicative loss rate ranges provided by the supervisors for specific loan categories.

• The estimates were adjusted by granular, bank- specific information on factors, such as past perfor-mance, portfolio composition, origination vintage, borrower characteristics, geographic distribution, international operations, and business mix, to

lender forbearance are discovered, through loss rec-ognition of unviable loans via additional provision-ing (which flows through profits to capital). This step would be taken ahead of the stress test, which would then provide an estimate of potential addi-tional capital needs under hypothetical adverse sce-narios (Appendix 14.2).

In contrast to Ireland and Spain, where variants of more comprehensive AQRs were conducted, the US stress test ap-plied a lower- intensity substitute. Supervisors addressed the

TABLE 14.9

Crisis Stress Tests: Disclosure ScorecardFramework Disclosure by Stress Test

Component Element United States European Union Ireland Spain

SCAP 2009

CEBS 2009

CEBS 2010

EBA 2011

PCAR 2011

FSAP 20121

TD 2012

BU 2012

Process Design Methodology Model(s) 2 Assumptions Macroeconomic scenarios 3 Loan loss assumptions Market assumptions P&L assumptions Behavioral assumptions (incl.

capital action as relevant)

Implementation Results System aggregate

Loan losses 4 By type Other portfolio losses 4 Impact on P&L Capital ratios Capital components (incl.

RWA)

Capital shortfall (incl. buffer) Capital action (incl.

government support measures as relevant)

Individual bank Loan losses 5 By type 5 Other portfolio losses 5 Impact on P&L Capital ratios Capital components (incl.

RWA)

Capital shortfall (incl. buffer as relevant)

Capital action (incl. government support measures as relevant)

Sources: Central Bank of Ireland; European Banking Authority; Federal Reserve; IMF; Roland Berger; and Oliver Wyman.Note: BU = bottom up; CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; FSAP = Financial Sector Assessment Program; incl. = including; PCAR = Prudential Capital Assessment Review; P&L = profit and loss; SCAP = Supervisory Capital Assessment Program; RWA = risk-weighted asset; TD = top down.1Included for completeness only—not intended as a crisis stress test; surveillance stress testing exercise was conducted in a crisis environment.2Some models published.3Very limited information disclosed. 4Combined amount provided.5Only loss rates provided.



and the type of loan and impact on the profit- and- loss ac-count. They also took into account banks’ proposed capital actions. The stock prices of financial firms appeared to ben-efit from the renewed transparency, having trended upward from late 2011, while volatility continued to moderate (Fig-ure 14.2). From 2013 onward, DFA stress tests are imple-mented alongside the CCAR (each with different capital action assumptions). The former requires annual and mid- cycle supervisory stress tests for systemically important fi-nancial institutions and the publication of those results.

The stress scenarios for the US supervisory stress tests were appropriately more onerous than that applied in the SCAP. The CCAR 2011 projected an adverse growth shock of 1.4 standard deviations from the projected baseline, while both the 2012 and 2013 stress tests assumed adverse growth scenarios of 2.5 standard deviations (Figure 14.7). In the DFA mid- cycle exercise, each bank develops its own baseline, adverse, and severely adverse scenarios to best re-flect its individual operations and risks; the banks are then required to publish the results of their respective severely ad-verse scenarios to help “promote market discipline and facili-tate an understanding of the financial conditions and risks” (Board of Governors of the Federal Reserve System 2013b). The supervisors use the stress test results to require banks to calibrate their proposed capital actions to ensure that they strengthen their capital positions.

While transparency of stress tests is critical in a crisis, its costs may be more subtle during normal times and may re-quire trade- offs. On the one hand, it would reduce opacity and instill market discipline. It could also be preemptive in terms of reducing uncertainty surrounding any public stress tests in future crises if markets get used to expecting that any adverse finding will entail appropriate follow- up action(s). On the other hand, as Schuermann (2013) observes, it could encourage banks to try and recreate the supervisory models rather than trace out their own risk profiles; or as noted by Goldstein and Sapra 2012, encourage banks to hold loan portfolios that generate good performance to pass the test, but which may not be beneficial for them in the longer term, lead to overreaction by markets ex post, or deter speculators from trading on their own views and market information, thus hampering the usefulness of that information for regula-tory purposes.

Follow- up stress tests were also implemented by Euro-pean supervisors. Given that Europe’s banking systems were not yet out of the woods, those tests provided opportunities for the authorities to improve upon previous exercises and to solidify previous gains made in regaining market confi-dence. Another round of EBA stress tests of EU banking sys-tems was undertaken in 2014, the fourth since its introduction in 2009, and since then, in 2016 and 2018. The ECB commenced its Comprehensive Assessment in 2014, comprising two main pillars: (1) an AQR; and (2) a stress test performed in close cooperation with the EBA. These may be carried out either regularly or on an ad hoc basis.

benchmark indicative loan loss parameters (Board of Governors of the Federal Reserve System 2009a).

• Reviews of the SCAP submissions by banks were subsequently conducted by experts in accounting and asset pricing and incorporated inputs from on- site supervisors.

Markets were reassured, and banks were able to recapital-ize and strengthen their balance sheets. The virtuous circle took hold as was the goal of the SCAP (Hirtle, Schuermann, and Stiroh 2009). The largest banks could confidently con-tinue to lend with the knowledge that the SCAP buffer would be adequate under adverse conditions, thus support-ing economic recovery, and consequently, the banks’ own asset quality. A possible reason for the market’s acceptance of the substitute to the AQR could be the credibility of the au-thorities’ review procedures and possibly perceptions of more reliable data quality in the first place.

One of the main shortcomings of the EU- wide stress tests had been the lack of any prior validation of banks’ asset quality. The EBA had recommended that national supervi-sors conduct AQRs on major EU banks ahead of the 2014 EU stress testing exercise, with the objective of reviewing asset classifications and valuations to help dispel concerns over deteriorating asset quality (EBA 2013). The exercise was coordinated with the planned balance sheet assessment of the Single Supervisory Mechanism (SSM) that was con-ducted by the ECB, in terms of its methodology and timing. The SSM exercise consisted of a comprehensive review of the banks that fell under the direct supervision of the ECB (see Constâncio 2013).

Follow- Up Stress Test(s)

Follow- up stress tests have been useful in consolidating the gains from crisis stress tests. In some cases, the former have evolved since their introduction during the crisis, with greater stringency and improved disclosure in some areas (Table 14.10). In the United States, follow- up supervisory stress tests to the SCAP have been conducted every year since. Separately, two other EU- wide crisis stress tests were conducted after 2009 in efforts to regain market confidence in the region’s banking system.

The US supervisory stress tests have taken disclosure to another level and may yet set a new benchmark for market expectations. The supervisors implemented the CCAR in 2011 but chose to keep the results confidential, although an overview of the exercise and the stress scenario were pub-lished (Board of Governors of the Federal Reserve System 2011a). It coincided with a weakening in the financials stock index (Figure 14.2). Following the low- profile CCAR 2011 exercise, information on the CCAR 2012 and CCAR 2013, as well as the DFA 2013 stress tests, was published in detail (Board of Governors of the Federal Reserve System 2011b, 2012a, 2012b, 2013a, 2013c). The disclosures included the frameworks, assumptions, methodologies, and bank- by- bank results of capital ratios; projected losses by portfolio;


Li Lian Ong and C


TABLE 14.10

European Union and the United States: Evolution of Publicized Stress Testing ExercisesJurisdiction Exercise Objective System Coverage Risk Horizon (RH) Growth Shock Scenarios Capital Standards Disclosure

Adverse

Number of Banks

Disclosed

Percent of

System Assets

Actual For Comparison

over Two-Year

Period

Std. Devn. (SD)

Calculation Period

SD of Two-Year Growth

Rate over Period

SD of Shock

over Two Years of

RH

Definition Metric(s) (Ratio)

Hurdle Rate(s)

Publica-tion of Results

Estimated Losses

(Billions)

United States

SCAP 2009 Crisis 19 67 2009–10 2009–10 1979–2008 2.9 1.0 Federal Reserve’s risk-based capital adequacy guidelines

Tier 1 common capital

Tier 1 capital

4

6

Yes USD599.2

European Union

CEBS Stress Test 2009

Crisis management

22 60 2009–10 2009–10 1979–2008 2.0 1.9 CRD Tier 1 capital 6 No n/a

European Union

CEBS Stress Test 2010

Crisis management

91 65 2010–11 2010–11 1980–2009 2.5 1.3 CRD Tier 1 capital 6 Yes EUR565.8

European Union

EBA Stress Test 2011

Crisis management

90 65 2011–12 2011–12 1981–2010 2.7 1.5 CRD Core Tier 1 5 Yes EUR400.0

European Union

EBA Capital Exercise 2011

Crisis management

65 (excl. GRE 6)

n/a Position as of Sep 2011

— — — — CRD Core Tier 1 9 Yes —

United States

CCAR 2011 Supervisory 19 65 2010:Q4– 2013:Q4

2011–12 1980–2009 3.4 1.4 Federal Reserve’s risk-based capital adequacy guidelines


Tier 1 capital

Total risk-based capital

Tier 1 leverage

5

6

8

3 or 4

No n/a

United States

CCAR 2012 Supervisory 19 67 2011:Q4– 2014:Q4

2011:Q4– 2013:Q3

1981–2010 3.5 2.5 Federal Reserve’s risk-based capital adequacy guidelines


Tier 1 capital


Tier 1 leverage

5

6

8

3 or 4

Yes USD534.0

(continued)


Credibility and C


Jurisdiction Exercise Objective System Coverage Risk Horizon (RH) Growth Shock Scenarios Capital Standards Disclosure

Adverse

Number of Banks Disclosed

Percent of System Assets

Actual For Comparison over Two-Year

Period

Std. Devn. (SD) Calculation Period

SD of Two-Year Growth Rate over Period

SD of Shock over Two Years

of RH

Definition Metric(s) (Ratio) Hurdle Rate(s) Publication of Results

Estimated Losses (Billions)

United States

DFA 2013 Supervisory 18 >70 2012:Q4– 2015:Q4

2012:Q4– 2014:Q3

1982–2011 3.5 2.5 Federal Reserve’s risk-based capital adequacy guidelines


Tier 1 capital


Tier 1 leverage

5

6

8

3 or 4

Yes USD462.0

United States

CCAR 2013 Supervisory 18 >70 2012:Q4– 2015:Q4

2012:Q4– 2014Q3

1982–2011 1979–2008

3.5 2.5 Federal Reserve’s risk-based capital adequacy guidelines


Tier 1 risk-based capital


5

6

8

Yes n/a

Sources: European Banking Authority; Federal Reserve; IMF; World Economic Outlook; and authors’ calculations. Note: Industry consensus growth forecasts applied for the Supervisory Capital Assessment Program (SCAP) exercise are the average of Consensus Forecasts, Blue Chip Economic Indicators, and Survey of Profes-sional Forecasters. No specific numbers are provided for the Comprehensive Capital Analysis and Review (CCAR) 2011 and 2012 baseline growth forecasts; identical industry sources, as SCAP are assumed. All US-domiciled banking organizations are required to compute risk-based capital requirements using the regulatory capital definition (general-risk-based capital rules, Basel I); none had entered a transitional floor period for risk-weighted assets as of 2011. The US Tier 1 leverage minimum is 3 percent for banks with a composite Bank Holding Company Rating System rating of “1” and for those that have implemented the Board’s risk-based capital measure for market risk; the minimum is 4 percent for all other banks. CCAR 2012 also applies Basel III framework calculations, fully phased. The Dodd-Frank Wall Street Reform and Consumer Protection Act (DFA) and CCAR are closely related, but with some important differences. The projections of pretax net income from the DFA exercise are used as direct inputs to the CCAR. The primary difference between the two is the capital action assumptions: the Federal Reserve uses a standardized set of capital action assumptions for the DFA; in contrast, bank holding companies’(BHCs’) planned capital actions are incorporated in the CCAR to project poststress capital ratios. CRD = capital requirement directive; CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; GRE = Greece; n/a = not applicable.




• Foremost is that macroprudential stress tests should assume the “going concern principle.” In other words, banks are assumed to operate as going con-cerns indefinitely and do not have to realize lifetime losses on their asset portfolios. This means that the banks are assumed to have the ability to hold loans to maturity, and stress test valuations are focused on projected cash- flow credit losses related to borrowers’ failure to meet their obligations, rather than on their liquidation values (see Bernanke 2009). Since the re-sults of crisis stress tests are used to help identify banks that may need to be restructured, standardiza-tion of scenarios and some key assumptions are nec-essary during this phase.

• Any required restructuring after the initial crisis stress test would be a more bespoke exercise. At that stage, a thorough assessment of the identified banks’ books prior to any recapitalization would be necessary. It would typically entail the recognition of valuation losses (for example, foreclosed real estate holdings or tax credits) or additional loan losses, which would also include some projections of future losses under stress, to determine an adequate capital buffer. Moreover, the banks’ noncore assets may have to be realized to-ward the cost of the restructuring effort, which could include selling off investment portfolios in their re-spective banking books at significant haircuts.

The Spain case represents a good example of how stress test numbers could differ significantly from estimated re-structuring costs (see Lister and Goodman 2012). The Fund for Orderly Bank Restructuring (Fondo de Reestructuración Ordenada Bancaria) noted three items on savings banks’ bal-ance sheets that were not captured in the crisis stress tests (see Alba 2011), namely: (1) compensation due for breach of in-surance contracts, (2) the fall in dividends from equity hold-ings, and (3) differences between deferred tax credits and realized tax credits. The Fund for Orderly Bank Restructur-ing’s explanation was that the nationalized institutions tend to be subject to more strenuous stress tests on these risk fac-tors, as they are required by the competition authorities to sell off their industry equity stakes and must mark them to market, while others have the option of keeping them on the books and are not required to recognize similar losses.

6. CONCLUDING REMARKSStress tests became an important instrument in supervisors’ crisis management toolkits during the global financial crisis. They are based on the concept of using a microprudential exercise for addressing macroprudential risks through im-proved transparency and disclosure. Introduced by the US authorities through the very high- profile SCAP in 2009, cri-sis stress tests have since been used by other jurisdictions with varying outcomes. The impact and case study analyses em-ployed in this chapter suggest that the design of particular elements of a stress test is critical if it is to be used for systemic

Liquidity Stress Test

Liquidity stress testing has become an important riskmanagement tool following the manifestation of unprece-dented liquidity shocks to the global banking system during the crisis. However, the earlier crisis stress tests had eschewed liquidity risk and focused on solvency risk instead. Schuer-mann (2012) observes that the “dynamism” of liquidity po-sitions, which are subject to rapid change, means that any snapshot at a particular point in time is unlikely to be infor-mative by the time of disclosure. The positive outcomes of some of the solvency stress tests suggest that markets did not necessarily require supporting liquidity tests. For example:

• Liquidity risk has not been specifically assessed as part of the EBA stress testing exercises. However, a confidential thematic review of liquidity funding risks was initiated in the first quarter of 2011 to as-sess banks’ vulnerability in relation to liquidity risk. The EBA 2011 solvency stress test analyzed the evo-lution of the cost of funding connected to the spe-cific financial structure of the banks in question, in particular, the impact of increases in interest rates on assets and liabilities, including that of sovereign stress on banks’ funding costs.

• Likewise, liquidity stress tests were not conducted in Spain’s case. However, the funding costs in the solvency stress tests were assumed to increase with the proposed scenarios for the solvency stress tests.

• Ireland’s Prudential Liquidity Assessment Review (PLAR) 2011 had been the only crisis liquidity stress test implemented. It covered the four PCAR banks. The exercise set bank- specific funding targets consis-tent with Basel III and other international measures of stable, high- quality funding (CBI 2011). The PCAR 2011 specified its constraints and parameters for funding costs and access to funds in line with the PLAR.

That said, it is unclear that supporting crisis liquidity stress tests would not have enhanced the solvency exercises, especially in Europe. Funding conditions for banks in the region remain impaired and the evidence suggests that mar-ket sentiment toward the banking sector and sovereigns only improved following the introduction of the LTRO and Out-right Monetary Transactions facilities (Figure 14.3).

5. COMPARING CRISIS STRESS TEST RESULTS WITH RESTRUCTURING COSTSThere has been much confusion over the divergences be-tween the capital shortfall of a bank estimated by a crisis stress test and its eventual recapitalization needs from any actual restructuring. In reality, the two exercises should not be expected to yield the same capital number given that they operate under vastly different assumptions. Rather, restructuring costs should be higher for the following reasons:



implementation may well require a dose of good for-tune, for example, in areas such as the actual health of banks, the timing of the exercise, market condi-tions, and public receptiveness to the disclosures (see Dudley 2011).

Ultimately, the lessons learned from this study suggest that country authorities must be fully committed if they are to undertake a crisis stress test. They must have a clear ob-jective and take action once valuations have fallen to certain levels. At that stage, and before any crisis of confidence be-comes firmly entrenched, they must be prepared to trans-parently conduct a thorough examination of their banking system, take necessary follow- up action(s) based on the findings, and have the requisite resources to back them, if the exercise is to serve its purpose of improving sentiment toward the banking system. Otherwise, the effort would likely backfire and exacerbate the loss in market confidence, with potentially devastating consequences for the real econ-omy. Many of the design features identified in this chapter are also relevant to confidential supervisory stress tests un-der crisis conditions, except perhaps for the transparency considerations.

Crisis stress tests may also be heralding a new era in trans-parency. Prior to the global financial crisis, supervisory stress tests were conducted in utmost confidentiality. The very public US crisis stress test was succeeded by supervisory stress tests that have since provided similar levels of disclosure. The EU has also continued to run and publish the results of its bank stress tests on a biennial basis. Opinion is divided as to the desirability of unfettered transparency of stress tests during normal times, but the bar for disclosure had been set high and markets appear to have become used to similar standards postcrisis.

crisis management. Moreover, an appreciation of certain con-cepts and nuances is necessary if the tool is to be applied con-structively and the results are to be properly interpreted.

Stress testing remains an art rather than a science, where expert judgment is indispensable. However, the use of stress tests for systemic crisis management has added other dimen-sions to this art form, notably:

• The public nature of crisis stress tests means that they must be designed to withstand intense scrutiny. Therefore, certain elements of the design, such as the timing of the test, its governance, the objective of the exercise, the proposed action plan to address the find-ings, and the nature of disclosure may necessarily have to be executed differently from what would be typical in normal supervisory stress tests.

• Other elements of crisis stress tests must be sufficiently rigorous so that the results are convincing. These in-clude the scope of coverage and the scenario design, although the latter need not necessarily be complex.

• Crisis stress tests also require the support of other activities to enhance their credibility. Specifically, AQRs are vital for the reliability of the inputs, while follow- up stress tests to update markets on develop-ments are important for consolidating previous gains. Separate liquidity stress tests to complement the solvency ones are also increasingly being employed, although not all are published.

• Political economy could play a key role in determin-ing the effectiveness of crisis stress tests. Given the potential economic and reputational implications of the findings, the design of these tests could be influ-enced by political economy considerations.

• Finally, it would be remiss to discount the impor-tance of luck in any crisis stress test. Its successful


Appendix 14.1.Case Studies of Crisis Solvency

Stress Tests: United States, European Union, Ireland, and Spain


Credibility and C



Crisis Stress Tests: Features of DesignFramework Application to Stress Test

Component ElementDesign Feature

United States European Union (EU) Republic of Ireland Spain

Supervisory Capital Assessment Program (SCAP)

Committee of European Banking Supervisors (CEBS) 2009



Prudential Capital Assessment Review (PCAR) 2011

IMF Financial Sector Assessment Program (FSAP) 20121

Top-Down (TD) 2012 Exercise

Bottom-Up (BU) 2012 Exercise

Effectiveness — Financials stock index stabilizes/improves, the index returns, volatility falls, and the sovereign CDS spread stabilizes or narrows.

• Yes—The financials stock index never again fell below the level recorded at the time the stress test results were announced; volatility dropped sharply following the announce-ment of results.

• No • No • No • Yes, the financials stock index rose sharply, the volatility of returns dropped, and the sovereign CDS spread narrowed significantly after the publication of the IMF’s Third Review in September 2011 helped give credence to the exercise.

• Not applicable—The FSAP stress test was not intended as a crisismanagement exercise.

• No • Yes—The financials stock stabilized around the level recorded at the time the stress test results were announced, volatility dropped sharply, and the sovereign CDS spread narrowed significantly.

Timing of exercise

— Stress test is conducted sufficiently early to arrest the decline in confidence

• The exercise was conducted 24 months after the peak of the financials stock index, while the index was still falling.

• The exercise was conducted 23 months after the peak of the financials stock index, while the index was still falling.

• The results were announced nine months following the CEBS 2009 results—after the Ireland and Greece bailouts provided some support for market sentiment— despite some loss in credibility.

• The results were announced 12 months after the CEBS 2010 results, but the exercise had lost significant credibility with markets by that stage.

• The exercise was conducted 49 months after the peak of the financials stock index, after almost all market value had been lost.

• Not applicable—FSAPs to each S-25 country are conducted mandatorily once every five years; the Spain FSAP took place 60 months after the peak of the financials stock index.

• The exercise was conducted 64 months from the peak of the financials stock index, after the supervisors had lost significant credibility with markets.

• The exercise was conducted three months after the TD stress test results, of which markets were skeptical, given the lack of information on individual banks.

• The index had fallen by 68 percent from the peak and the price-to-book ratio was 0.99 as at the end of 2008.



• The index had fallen by 70 percent from the peak at the time the FSAP report was published in June.

• The index had fallen by 61 percent from the peak and the price-to- book ratio was 0.26 as at the end of 2011.

(continued)


Li Lian Ong and C














Governance Oversight Stress test is overseen by an independent third party.

• No—The stress test was overseen by the authorities.

• Yes—The stress test was overseen and coordinated by the CEBS in cooperation with the EC and the ECB.

• Yes—The stress test was overseen and coordinated by the CEBS in cooperation with the EC and the ECB.

• Yes—The stress test was overseen and coordinated by the EBA in cooperation with the EC, the ECB, and the ESRB.

• No—The stress test was overseen by the CBI.

• Not applicable.

• Yes—The stress test was overseen by a Steering Committee comprising the authorities, supported by an advisory panel consisting of the Troika, Banque de France, and the Dutch National Bank.

• Yes—The stress test was overseen by a Strategic Coordination Committee consisting of the authorities, the Troika, and the EBA and in consultation with an Expert Coordination Committee comprising the same membership.

Stress tester(s) Stress test is conducted by an “indepen-dent” third party.

• No—The stress test was conducted by the authorities (Federal Reserve, FDIC, and OCC).

• No—The stress test was conducted by national supervisory authorities.



• No—The stress test was conducted by the CBI with BlackRock Solutions providing forecast losses; Boston Consulting Group provided oversight and challenge to BlackRock’s work.

• No—The stress test was conducted by the BdE, in collaboration with the IMF.

• Yes—The stress test was conducted by Oliver Wyman and Roland Berger.

• Yes—The stress test was conducted by Oliver Wyman, with inputs from auditors and real estate valuation companies, all coordinated by the Boston Consulting Group.

(continued)


Credibility and C














Scope Approach Stress test approach is BU.

• Yes—BU and TD mix, with authorities providing macroeconomic scenarios and guidance on the estimation of loan loss parameters.

• Yes— Constrained BU.

• Yes— Constrained BU (of cross- border groups) and simplified stress tests, based on national supervisors and reference parameters provided by the ECB for less complex institutions, with macroeconomic and sovereign shock scenarios and parameters developed by the ECB.

• Yes— Constrained BU (of cross- border groups) and simplified stress tests, based on national supervisors and reference parameters provided by the ECB for less complex institutions, with macroeconomic and sovereign shock scenarios and parameters developed by the ECB.

• No—Constrained TD, relying on BlackRock’s assessment of forecast losses and incorporating much of the methodology and parameters used for the EBA stress test.

• No— Constrained TD, with the authorities running both IMF and BdE models, applying growth scenarios and assumptions agreed on with the IMF.

• No—TD of individual banks with growth scenarios and some guidelines provided by steering committee comprising the Troika, and internal and external agencies.

• No—“BU” exercise comprising TD of individual banks, with growth scenarios and some guidelines provided by steering committee comprising the Troika, internal agencies, and other EU counterpart agencies.

• Process includes supervisors applying independent quantitative methods using firm-specific data to support their assessments of banks’ submissions.

• Process includes a peer review of results and challenging process, and extensive cross-checks by the CEBS.

• Process includes consistency checks by the EBA, a multilateral review and TD analysis by the EBA and the ESRB with ECB assistance.

• Process includes close consultation with an Expert Coordination Committee comprising the Troika, the EBA, and the authorities.

Coverage Stress test covers at least the systemically important banks and the majority of banking system assets

• Yes—19 BHCs each with assets >$100 billion.

• Yes— 22 major EU cross-border banking groups.

• Yes— 91 banks, comprising major EU cross-border banking groups and a group of additional, mostly larger credit institutions.

• Yes— 90 banks, comprising major EU cross-border banking groups and a group of additional, mostly larger credit institutions.

• Yes— All Irish domestically owned commercial banks (four).

• Yes— Commercial banks and intervened savings banks (13), including the largest and problem banks.

• Yes— Commercial banks and intervened savings banks (14), including the largest and problem banks.

• Yes—14 largest merged banking groups, disaggregated into 17 following stress test.

(continued)


Li Lian Ong and C














Scope (continued)

Coverage(continued)

Stress test covers at least the systemically important banks and the majority of banking system assets

• Yes—2/3 of banking system assets.

• Yes— 60 percent of the EU banking system assets.

• Yes— 65 percent of the EU banking system assets and at least 50 percent of assets of each national banking sector.

• Yes— 65 percent of the EU banking system assets and at least 50 percent of assets of each national banking sector.

• Yes—Approxi-mately 80 percent of banking system assets including foreign subsidiaries.

• Yes—88 percent of banking system assets excluding foreign branches.

• Yes— Approximately 90 percent of banking system assets excluding foreign branches.

• Yes— Approximately 90 percent of banking system assets excluding foreign branches.

Scenario design Scenarios Stress test applies large scenario shocks (2 std. devn. or larger)

• No—One baseline and one adverse (equivalent to 1 std. devn. of cumulative shock from baseline over two years).

• No—One baseline and one adverse (1.9 std. devn. of cumulative shock from baseline over two years).



• No—One baseline and one adverse (consistent with the EBA 2011 stress testing exercise).

• No—One baseline and two adverse (more adverse scenario is 1 std. devn. of cumulative shock from baseline over two years).

• No—One baseline and two adverse (first two years identical to FSAP adverse scenario, plus a third year of negative growth).

• No—One baseline and one adverse (first two years identical to FSAP adverse scenario, plus a third year of negative growth).

Risk factors Stress test applies shocks to key risk factors

• Yes—Credit risk: 12 categories of loans stress tested.

• Credit risk: Stress tested but little information available.

• Yes—Credit risk: five main portfolios stress tested.

• Yes—Credit risk: eight categories of loans stress tested; sovereign exposure in AfS banking book treated as credit risk.

• Yes—Credit risk: five categories of loans stress tested.

• Yes—Credit risk: five categories of loans plus foreclosed assets stress tested.

• Yes—Credit risk: six categories of loans plus foreclosed assets stress tested.

• Yes—Credit risk: six categories of loans plus five segments of foreclosed assets stress tested.

(continued)


Credibility and C














Scenario design (continued)

Risk factors (continued)

Stress test applies shocks to key risk factors (continued)

• Yes— Market risk: Investment (and trading securities portfolios for firms with trading assets >$100 billion) in the trading book, AfS and HtM in the banking book stress tested.

• No—Market risk: Sensitivity analysis of trading book only.

• No—Market risk: Sovereign and financial institution exposures stress tested, but in the trading book only.

• No—Market risk: Sovereign and financial institution exposures stress tested, but in the trading book only.

• Market risk: Securities portfolio stress tested but insignificant; no sovereign haircut.

• Market risk: Sovereign portfolio stress tested, but in the trading book and the AfS banking book only.

• Market risk: Sovereign portfolio not stress tested, but ECB LTRO facility in place.

• Market risk: Sovereign portfolio not stress tested, but ECB LTRO facility in place.

Assumptions Macroeco-nomic scenarios are standardized across banks

• Yes—Baseline based on industry consensus forecasts; adverse provided by authorities.

• Yes— Based on collabo ration among the CEBS, the EC, and the ECB.

• Yes—Based on collaboration among the CEBS, the EC, and the ECB.

• Yes—Based on collaboration among the EBA, the EC, and the ECB/ESRB.

• Yes—Based on the EBA scenarios.

• Yes—Agreed upon between IMF staff and authorities.

• Yes—Based on FSAP scenarios.

• Yes—Based on FSAP scenarios, extended to a three-year horizon.

Behavioral assumptions are standardized across banks

• No—Banks’ own. • Yes—Bench-mark risk parameters provided by the ECB; process and guidelines provided by the CEBS.

• Yes—Range of assumptions provided by the CEBS, in cooperation with the ECB, the EC, and national supervisory authorities.

• Yes—Range of assumptions provided by the EBA, in cooperation with the EC, the ECB, the ESRB, and national supervisory authorities.

• Yes—Assumptions largely based on the EBA stress test.

• Yes—Range of assumptions provided by IMF staff.

• No—Assump-tions provided by stress testers except for macroscenarios.

• No—Assumptions except for macroscenarios provided by stress tester with some guidance from steering committee.

Capital standards

. . . Stress test applies very high hurdle rate(s) (CT1 > 6 percent)

• No—Hurdle rates of T1 capital of 6 percent and T1 common capital of 4 percent for both baseline and adverse scenarios.

• No—Hurdle rate of T1 capital of 6 percent for both baseline and adverse scenarios.

• No—Hurdle rate of T1 capital of 6 percent for both baseline and adverse scenarios.

• No—Hurdle rate of CT1 capital of 5 percent for both baseline and adverse scenarios.

• Yes—Hurdle rates of CT1 capital of 10.5 percent under the baseline scenario and 6 percent under the adverse scenario.

• Yes—Hurdle rates of total capital of 8 percent, T1 capital of 6 percent, and CT1 capital of 4 and 7 percent for baseline and both adverse scenarios.

• Yes—Hurdle rates of CT1 capital of between 6 and 9 percent.

• Yes—Hurdle rates of CT1 capital of 9 percent under the baseline scenario and 6 percent under the adverse scenario.

(continued)


Li Lian Ong and C














T r a n s p a r e n c y

Objective and action plan

Objective Stress test is associated with a clear and resolute objective

• Yes—Stated objective was specifically to assess the capital needs of banks that would provide a buffer against higher than generally expected losses and still be able to lend to creditworthy borrowers should such losses materialize.

• No—Stated objective was to carry out an EU-wide forward- looking stress testing exercise on the aggregate banking system.

• Yes—Stated objective was to provide policy information for assessing the resilience of the system to possible adverse economic developments and to assess the ability to absorb possible shocks on credit and market risks, including sovereign risks.

• Yes—Stated objective was to assess the resilience of individual institutions and the system to hypothetical stress events under certain restrictive conditions.

• Yes—Stated objective was to determine the capital resources of domestic banks under a given stress scenario, in order to calculate the cost of recapitalization required to meet central bank-imposed requirements.

• No—FSAP stress tests are conducted as part of the overall stability analysis of a financial system and to facilitate policy discussions on crisis preparedness. This objective was not stated explicitly.

• No—Stated objective was broadly to increase transparency and dispel doubts over the valuation of bank assets.

• Yes—Stated objective was to adhere to the MoU approved by the Eurogroup on July 20, 2012, which required the estimation of capital needs as an essential element of the roadmap for the recapitaliza-tion and restructuring of the banking system.

Follow-up action(s)

Stress test is associated with clear follow-up action(s) by manage-ment/ authorities to address findings as necessary.

• Yes—Banks needing to augment capital buffers were required to develop a detailed capital plan to be approved and implemented within six months.

• No— National authorities were responsible for any follow-up to the exercise.

• No—National authorities were responsible for any follow-up to the exercise.

• Yes—Banks showing capital shortfalls were required to present their plans to restore capital position and to implement remedial measures by the end of the year.

• Yes—Recapitaliza-tion was required based on loan loss projections, along with further calculations of prospective income, expenditure, and deleveraging plans.

• No—FSAP stress tests are surveillance purposes and to facilitate policy discussions; they typically do not require management action.

• No • Yes—Recapitaliza-tion/restructuring based on stress test findings required by the MoU with the Eurogroup.

Financing backstop

Stress test is provided with an explicit financial backstop to support the necessary follow-up action(s).

• Yes—Capital Assistance Program under the Troubled Asset Relief Program.

• No—Not for the region as a whole.



• Yes—Already in a crisis program with the Troika.

• No • No • Yes—The ESM facility per the MoU with the Eurogroup.

(continued)


Credibility and C














T r a n s p a r e n c y

Disclosure of technical details

Design, methodol-ogy and implemen-tation


• Yes—Detailed information provided on stress test design, methodology, and implementation.

• No—Minimal information on exercise was provided.

• Yes—Detailed information provided on stress test design, methodology, and implementa-tion.

• Yes—Detailed information provided on stress test design, methodology, and implementa-tion.

• Yes—Detailed information provided on stress test design, methodology, and implementation

• Yes—Detailed information provided on stress test design and methodol-ogy.

• Yes—Some information provided on stress test design, methodology, and implementa-tion

• Yes—Detailed information provided on stress test design, methodology, and implementation.

Model(s) Stress test discloses information

• No—Information on banks’ stress test models not disclosed, but projections subjected to detailed review and assessment by supervisors.

• No—Infor-mation on banks’ stress test models not disclosed, but results subjected to peer review and challenging process.

• No—Informa-tion on banks’/national supervisory authorities’ models not disclosed, but results subjected to peer review and challenging process.

• No—Information on banks’/ national supervisory authorities’ models not disclosed, but results subjected to peer review and challenging process.

• No—Information on banks’ and third-party models not disclosed, but supervisory challenges and independent assessment undertaken.

• Yes—Informa-tion provided on IMF and Bank of Spain stress test models; results from two models cross- validated.

• Some information provided on third-party stress test models.

• No—Information on third-party stress test models not disclosed but modeling process shared and discussed with representatives of the steering committee.

Details of assump-tions


• Yes—High level information on macroeconomic assumptions; detailed information on loan loss assumptions.

• No—Very limited information provided on macroeco-nomic assumptions.

• Yes—Detailed information on macroeconomic and market assumptions.

• Yes—Detailed information on macroeconomic and market assumptions.

• Yes—Detailed information on macroeco nomic and P&L assumptions and loan losses.

• Yes—Detailed information on macroeco-nomic and other behavioral assumptions.

• Yes—Detailed information on macroeco-nomic, P&L, and loan loss assumptions/estimates.

• Yes—Detailed information on macroeconomic and P&L assumptions and loan loss estimates.

Bank-by-bank results

Stress test discloses detailed information

• Yes—Summary results disclosed at individual BHC level, including projected losses, capital components, and capital needs.

• No—Sum-mary results disclosed at system aggregate level.

• Yes— Summary results disclosed at system aggregate and individual bank levels.

• Yes—Summary results disclosed at system aggregate and individual bank levels, including projected losses, capital components, and capital needs.

• Yes—Detailed results disclosed at individual bank level, including projected losses, capital components, and capital needs.

• No— Summary results provided at aggregated groupwise (according to specific characteris-tics) and system aggregate levels.

• No—Summary results provided at system aggregate levels.

• Yes—Summary results disclosed at individual bank level, including projected loan losses, capital components, and capital needs.

(continued)


Li Lian Ong and C














Asset quality review (AQR)

. . . AQR is undertaken as input into stress test

• Yes— Lower-intensity, quantitative substitute.

• No • No • No • Yes—Sample loan review and independent audit conducted.

• No • No • Yes—Deep dive conducted, supported by real estate appraisers and independent audit.

Liquidity stress test

. . . Liquidity stress test accompanies solvency stress test

• No • No • No • No—But liquidity profile of banks assessed through a confidential specific thematic review for supervisory purposes.

• Yes—Prudential Liquidity Assessment Review.

• Yes • No • No

Follow-up stress tests

. . . Subsequent stress tests are conducted at regular intervals to maintain transparency

• Yes—CCAR 2011, 2012, 2013, and DFA 2013 by the authorities, designed in a similar manner to the SCAP.

• Yes—CEBS 2010 by the authorities.

• Yes—EBA 2011 by the authorities.

• Yes—EBA 2014 by the authorities.

• Undertaken in the context of the SSM’s 2014 Comprehensive Assesssment and EBA 2014.

• Not applicable— FSAPs to each S-25 country were conducted mandatorily once every five years.

• Yes—“BU” exercise as part of MoU with the Troika.

• Yes—Undertaken in the context of the SSM’s 2014 Comprehensive Assesssment and EBA 2014.

• Conducted EU Recapitalization Exercise requiring banks to strengthen their capital positions against sovereign debt exposures.

Sources: Banco de España (BdE); Central Bank of Ireland (CBI); European Banking Authority (EBA); Federal Reserve; IMF; and authors. Note: AfS = available for sale; BHC = bank holding company; CCAR = Comprehensive Capital Analysis and Review; CDS = credit default swap; CT1 = Core Tier 1; DFA = Dodd-Frank Wall Street Reform and Consumer Protection Act; EC = European Commission; ECB = European Central Bank; ESM = European Stability Mechanism; ESRB = European Systemic Risk Board; FDIC = Federal Deposit Insurance Corporation; HtM = hold to maturity; LTRO = Long-Term Refinancing Operation; MoU = memorandum of understanding; OCC = Office of the Comptroller of the Currency; P&L = profit and loss; S-25 = Systemic-25 jurisdictions; SSM = Single Supervisory Mechanism; Std. devn. = standard deviation; T1 = Tier 1.1Included for completeness only—not intended as a crisis stress test; surveillance stress testing exercise was conducted in a crisis environment.


Appendix 14.2.The Cost of Direct Recapitalization

versus Loss Recognition and Provisioning

A crisis stress test may be undertaken with or without a prior asset quality review (AQR). In the former, the data reported by banks are accepted as is; in the latter, an audit and possibly asset valuation exercise and data integrity and verification are per-formed to ensure that the data are “clean.” An AQR should be done if the quality of data is highly suspect such that any stress test result would be considered meaningless without it, which is typically the case in a crisis. AQRs tend to be more costly, not just from the actual cost of running the exercise itself, but because they usually require a cleaning up of the books before any stress test is conducted. As a simple example, a situation is assumed where the stress test shows total capital needs of €50 billion (with constant risk- weighted assets) in order to meet the required hurdle rate (Appendix Figure 14.2.1):

• In Example 1, no AQR is conducted and the existing data are used in the stress test:1. The stress test projects a fall in revenue of €10 billion.2. Expected loan losses are €20 billion.3. Net profit decreases from €40 billion to €10 billion.4. The projected capital shortfall is €10 billion and is addressed through a direct cash injection.

• In Example 2, an AQR is conducted first, followed by the stress test:1. Lender forbearance, loan misclassifications, and incorrect valuations are discovered, leading to additional required

provisioning of €30 billion.2. The stress test subsequently projects a drop in revenue of €10 billion.3. Loan losses are estimated at another €20 billion.4. The result is that the bank makes a net loss of €20 billion.5. The projected capital shortfall is now €40 billion.

Put another way, the higher cost of conducting an AQR is attributable to the “double hit” to the books. The findings of the AQR may require loss recognition in the profit- and- loss account first through accounting for additional provisions or inaccu-rate valuations, ahead of any additional recapitalization that may be required from a crisis stress test.


Credibility and C


Source: Authors.Note: AQR = asset quality review; P&L = profit and loss.

Appendix Figure 14.2.1 Accounting Entries: Loss Recognition, Provisioning, and Recapitalization (In billions of euro)

Various expensesProvisionProfit

1100

40

Various revenues 150 LoansLess provisionsLoans net of provisionsCash

24020

22020

Various liabilities 170

SharesRetained earningsProfit transferred from P&L

201040

240150 150 240

Various expensesProvision from stress testProfit

1102010

Various revenues reduced by stress test

140 1 LoansLess existing provisionsLess provision from stress testLoans net of provisionsCash reduced by drop in revenueCash injection for capital

2402020

2001010

220

Various liabilities 17023

2SharesRetained earningsProfit transferred from P&LAdditional capital injection

20101010

220

14

140 140

Expenses Revenues Assets Liabilities

Capital

Capital

Balance SheetProfit & Loss

Expenses Revenues Assets LiabilitiesBalance SheetProfit & Loss

Starting position

Example 1: Run stress test and increase capital to required €50 billion

Various expensesProvision from AQRProvision from stress test

1103020

Various revenues reduced by stress testLoss

140

20

2

4

LoansLess existing provisionsLess provision from AQRLess provision from stress testLoans net of provisionsCash reduced by drop in revenueCash injection for capital

240203020

1701040

220

Various liabilities 17013 1

3 SharesRetained earningsLoss transferred from P&LAdditional capital injection

2010

–2040

220

25

34

45

160 160

Capital

Expenses Revenues Assets LiabilitiesBalance SheetProfit & Loss

Example 2: Realize losses of €30 billion from AQR first and then run stress test and increase capital to €50 billion



———. 2009c. The Supervisory Capital Assessment Program: Over-view of Results. Washington, DC: Board of Governors of the Fed-eral Reserve System. https://www.federalreserve.gov/newsevents /pressreleases/bcreg20090507a.htm.

———. 2011a. Comprehensive Capital Analysis and Review: Objec-tives and Overview. Washington, DC: Board of Governors of the Federal Reserve System, March 18. http://www.federalreserve .gov/newsevents/press/bcreg/20110318a.htm.

———. 2011b. Federal Reserve System Comprehensive Capital Anal-ysis and Review: Summary Instructions and Guidance. Washing-ton, DC: Board of Governors of the Federal Reserve System, November 22. http://www.federalreserve.gov/newsevents/press /bcreg/20111122a.htm.

———. 2012a. Comprehensive Capital Analysis and Review 2012: Methodology and Results for Stress Scenario Projections. Washing-ton, DC: Board of Governors of the Federal Reserve System, March 13. http://www.federalreserve.gov/newsevents/press/bcreg /20120313a.htm.

———. 2012b. Comprehensive Capital Analysis and Review 2013: Summary Instructions and Guidance. Washington, DC: Board of Governors of the Federal Reserve System, November 9. http://www.federalreserve.gov/newsevents/press/bcreg/20121109b .htm.

———. 2013a. Comprehensive Capital Analysis and Review 2013: Assessment Framework and Results. Washington, DC: Board of Governors of the Federal Reserve System, March 14. http://www.federalreserve.gov/newsevents/press/bcreg/20130314a .htm.

———. 2013b. Dodd- Frank Act Mid- Cycle Stress Tests 2013: Sum-mary Instructions. Washington, DC: Board of Governors of the Federal Reserve System, May. https://www.federalreserve.gov /newsevents/pressreleases/bcreg20130513a.htm.

———. 2013c. Dodd- Frank Act Stress Test 2013: Supervisory Stress Test Methodology and Results. Washington, DC: Board of Gov-ernors of the Federal Reserve System, March 7. https://www .federalreserve.gov/newsevents/pressreleases/bcreg20130307a .htm.

———. 2018a. Dodd- Frank Act Stress Test 2018: Supervisory Stress Test Methodology and Results. Washington, DC: Board of Gov-ernors of the Federal Reserve System, June 28. https://www .federalreserve.gov/publications/2018- june- dodd-frank-act -stress-test-preface.htm.

———. 2018b. Federal Reserve Releases Results of Comprehensive Capital Analysis and Review (CCAR). Washington, DC: Board of Governors of the Federal Reserve System, June 28. https://www .federalreserve.gov/newsevents/pressreleases/bcreg20180628a .htm.

Federal Deposit Insurance Corporation and Office of the Comptroller of the Currency (Federal Reserve/FDIC/OCC). 2012. SR 12-7: Supervisory Guidance on Stress Testing for Banking Organizations with More Than $10 Billion in Total Consolidated Assets. Washing-ton, DC: Federal Reserve/FDIC/OCC, May. https://www.feder-alreserve.gov/supervisionreg/srletters/sr1207.htm.

Borio, Claudio, Mathias Drehmann, and Kostas Tsatsaronis. 2012. “ Stress- Testing Macro Stress Testing: Does It Live Up to Expectations?” BIS Working Paper 369, Bank for International Settlements, Basel, Switzerland, January. http://www.bis.org /publ/work369.htm.

Borio, Claudio, Bent Vale, and Goetz von Peter. 2010. “Resolving the Financial Crisis: Are We Heeding the Lessons from the Nordics”? BIS Working Paper 311, Bank for International Settlements, Basel, Switzerland. http://www.bis.org/publ/work311.htm.

REFERENCESAcharya, Viral V., Itamar Drechsler, and Philipp Schnabl. 2011.

“Pyrrhic Victory?—Bank Bailouts and Sovereign Credit Risk.” NBER Working Paper 17136, National Bureau of Economic Re-search, Cambridge, MA. http://www.nber.org/papers/w17136.

Ahmed, Enam, Andrea Appeddu, Melanie Bowler, Tomas Ho-linka, Juana Manuel Licari, Olga Loiseau- Aslanidi, and Zach Witton. 2011. “Europe Misses Again on Bank Stress Test.” Moody’s Analytics Regional Financial Review, July.

Alba, Miguel. 2011. “El FROB Reconoce Que el Test de Oliver Wyman se Quedó Corto: Detecta Ya Tres Nuevos Peligros.” Voz-populi, February 26. http://vozpopuli.com/ economia- y- finanzas /21932- el- frob- reconoce- que- el- test- de- oliver- wyman- se - quedo- corto-detecta-ya-tres-nuevos-peligros.

Angeloni, Chiara, and Guntram Wolff. 2012. “Sovereign Portfo-lios or Banks’ Location: What Channels Sovereign Risk in to Banking Systems?” Vox: CEPR’s Policy Portal. http://www .voxeu.org/article/ sovereign- debt-and-bank-risk-new-evidence.

Banco de España (BdE). 2013. Financial Stability Report. Madrid, Spain: Banco de España, May. http://www.bde.es/bde/en/sec ciones /informes/boletines/Informe_de_Estab/anoactual.

Bank Regulation and Supervision Agency (BRSA). 2002a. Bank Capital Strengthening Program. Ankara, Turkey: Bank Regula-tion and Supervision Agency, February 20. http://www.bddk .org.tr/WebSitesi/english/Reports/Other_Reports/Other_ Reports.aspx.

———. 2002b. Bank Capital Strengthening Program Progress Re-port. Ankara, Turkey: Bank Regulation and Supervision Agency, June 21. http://www.bddk.org.tr/WebSitesi/english /Reports/Other_Reports/Other_Reports.aspx.

Basel Committee on Banking Supervision (BCBS). 2008. Princi-ples for Sound Stress Testing Practices and Supervision: May 2009. Basel, Switzerland: Bank for International Settlements. https://www.bis.org/publ/bcbs155.htm.

———. 2011. “Global Systemically Important Banks: Assessment Methodology and the Additional Loss Absorbency Require-ment.” BCBS Publication 207, Bank for International Settle-ments, Basel, Switzerland, November. http://www.bis.org/publ /bcbs207.htm.

———. 2012. “A Framework for Dealing with Domestic Systemi-cally Important Banks: Final Document.” BCBS Publication 233, Bank for International Settlements, Basel, Switzerland, October. http://www.bis.org/publ/bcbs233.htm.

Bernanke, Ben S. 2009. “The Supervisory Capital Assessment Pro-gram.” Speech presented at the Federal Reserve Bank of Atlanta 2009 Financial Markets Conference, Jekyll Island, Georgia, May 11. http://www.federalreserve.gov/newsevents/speech/bernanke 20090511a.htm.

———. 2010. “The Supervisory Capital Assessment Program— One Year Later.” Speech presented at Federal Reserve Bank of Chicago 46th Annual Conference on Bank Structure and Competition, Chicago, IL, May 6. https://www.federalreserve .gov/newsevents/speech/bernanke20100506a.htm.

Board of Governors of the Federal Reserve System (Federal Re-serve). 2009a. Frequently Asked Questions— Supervisory Capital Assessment Program. Washington, DC: Board of Governors of the Federal Reserve System.

———. 2009b. The Supervisory Capital Assessment Program: Design and Implementation. Washington, DC: Board of Governors of the Federal Reserve System, April 24. https://www.federalreserve .gov/newsevents/pressreleases/bcreg20090507a.htm.


https://www.federalreserve.gov/newsevents/pressreleases/bcreg20090507a.htm.


http://www.federalreserve.gov/newsevents/press/bcreg/20110318a.htm.






http://www.federalreserve.gov/newsevents/press/bcreg/20121109b.htm.











https://www.federalreserve.gov/newsevents/pressreleases/bcreg20180628a.htm



https://www.feder-alreserve.gov/supervisionreg/srletters/sr1207.htm

https://www.feder-alreserve.gov/supervisionreg/srletters/sr1207.htm

http://www.bis.org/publ/work369.htm



http://www.nber.org/papers/w17136

http://vozpopuli.com/

http://www.voxeu.org/article/


http://www.bde.es/bde/en/sec

http://www.bddk.org.tr/WebSitesi/english/Reports/Other_Reports/Other_Reports.aspx.



http://www.bddk.org.tr/WebSitesi/english/Reports/Other_Reports/Other_Reports.aspx

http://www.bddk.org.tr/WebSitesi/english/Reports/Other_Reports/Other_Reports.aspx





http://www.bis.org/publ/bcbs233.htm

http://www.federalreserve.gov/newsevents/speech/bernanke







http://www.bde.es/bde/en/sec

https://www.federalreserve.gov/publications/2018-june-dodd-frank-act-stress-test-preface.htm




https://www.eba.europa.eu/-/ the- eba- details- the- eu- measures - to- restore-confidence-in-the-banking-sector.

———. 2011b. EU- Wide Stress Test Aggregate Report. London, UK: European Banking Authority, July 15. https://www.eba.europa .eu/ risk- analysis- and-data/eu-wide-stress-testing/2011/results.

———2011c. 2011 EU- Wide Stress Test: Methodological Note. London, UK: European Banking Authority, March 18. https://www .eba.europa.eu/-/ the- eba- publishes- details- of- its-stress-test -scenarios-and-methodology.

———. 2011d. Questions and Answers on the EBA 2011 EU- Wide Stress Test. London, UK: European Banking Authority, March 18. https://eba.europa.eu/ risk- analysis-and-data/eu-capital -exercise/2011.

———. 2011e. Results of Bank Recapitalization Plan. London, UK: European Banking Authority. December 8. http://www.eba .europa.eu/ risk- analysis-and-data/eu-capital-exercise.

Fama, Eugene F. 1970. “Efficient Capital Markets: A Review of Theory and Empirical Work.” Journal of Finance. 25 (2): 383–417.

Fox, Justin. “2009. First of All, It’s Not a Stress Test, It’s a SCAP…” Time, April 24. http://business.time.com/2009/04/24/ first- of - all- its- not- a-stress-test-its-a-scap.

Garicano, Luis. 2012. “Five Lessons from the Spanish Cajas Deba-cle for a New Euro- wide Supervisor.” Vox: CEPR’s Policy Portal, October 16. http://www.voxeu.org/article/ five- lessons- spanish - cajas-debacle-new-euro-wide-supervisor.

Goldstein, Itay, and Haresh Sapra. 2012. Should Banks’ Stress Test Re-sults Be Disclosed? An Analysis of the Costs and Benefits. Cambridge, Massachusetts: Committee on Capital Markets Regulation.

Helmore, Edward. 2008. “The New Sage of Wall Street.” Wall Street Journal, September 27. http://www.guardian.co.uk/books/2008 /sep/28/businessandfinance.philosophy.

Hirtle, Beverly, Til Schuermann, and Kevin Stiroh. 2009. “Macro-prudential Supervision of Financial Institutions: Lessons from the SCAP.” FRBNY Staff Report 409, Federal Reserve Bank of New York, New York, NY. https://www.newyorkfed.org /re search/staff_reports/sr409.html.

International Monetary Fund (IMF). 2008. “Hungary: Request for Stand- By Arrangement— Staff Report.” IMF Country Re-port 08/361, Washington, DC. http://www.imf.org/external /pubs/cat/longres.aspx?sk=22493.

———. 2011. “Ireland: Third Review under the Extended Arrange-ment.” IMF Country Report 11/276, Washington, DC. http://www.imf.org/external/pubs/cat/longres.aspx?sk=25223.

———. 2012a. “ Macro- Financial Stress Testing: Principles and Practices.” IMF Policy Paper, Washington, DC. http://www .imf.org/external/pp/longres.aspx?id=4702.

———. 2012b. “Spain: Financial System Stability Assessment.” IMF Country Report 12/137, Washington, DC. http://www .imf.org/external/pubs/cat/longres.aspx?sk=25977.

———. 2013a. “European Union: Publication of Financial Sector Assessment Program Documentation— Technical Note on Stress Testing of Banks.” IMF Country Report 13/68, Washing-ton, DC. http://www.imf.org/external/pubs/cat/longres.aspx?sk= 40396.

———. 2013b. “Key Aspects of Macroprudential Policy.” IMF Policy Paper, Washington, DC. https://www.imf.org/en /Publications/ Policy- Papers/Issues/2016/12/31/ Key-Aspects -of-Macroprudential-Policy-PP4803.

International Monetary Fund, Financial Stability Board, and Bank for International Settlements (IMF/FSB/BIS). 2009. Guidance to Assess the Systemic Importance of Financial Institutions, Markets and

Brunsden, Jim. 2012. “EBA Needs More Legal Powers for Stress Tests, Enria Says.” Bloomberg, September 19.

Campbell, Alexander. 2011 “EBA Stress Tests: Adverse Scenario Toughened Up.” Risk.net. March 18. http://www.risk.net / risk- magazine/news/2035333/ eba-stress-tests-adverse-scenario -toughened.

Central Bank of Ireland (CBI). 2011. The Financial Measures Pro-gramme Report. Dublin, Ireland: Central Bank of Ireland, March 31. https://www.centralbank.ie/publication/financial -measures-programme.

Claessens, S., C. Pazarbasioglu, L. Laeven, M. Dobler, F. Valen-cia, O. Nedelescu, and K. Seal. 2011. “Crisis Management and Resolution: Early Lessons from the Financial Crisis.” IMF Staff Discussion Note 11/05, International Monetary Fund, Washing-ton, DC. http://www.imf.org/external/pubs/cat/longres.aspx?sk= 24694.0.

Committee of European Banking Supervisors (CEBS). 2009a. CEBS’ Statement on Stress Testing Exercise. London, UK: Com-mittee of European Banking Supervisors, May. https://www .eba.europa.eu/-/ cebs- s-statement-on-stress-testing-exercise.

———. 2009b. “Results of the EU- Wide Stress Testing Exercise.” Press Release, Committee of European Banking Supervisors, London, UK, October 1. https://www.eba.europa.eu/-/ cebs- press - release- on- the- results- of- the-eu-wide-stress-testing-exercise.

———. 2010a. Aggregate Outcome of the 2010 EU- wide Stress Test Exercise Coordinated by CEBS in Cooperation with the ECB. Lon-don, UK: Committee of European Banking Supervisors, July 23. https://www.eba.europa.eu/ risk- analysis- and-data/eu-wide -stress-testing/2010/results.

———. 2010b. Questions & Answers: 2010 EU- Wide Stress Testing Exercise. Committee of European Banking Supervisors: London, UK, July 23. http://www.eba.europa.eu/ risk- analysis- and-data /eu-wide-stress-testing/2010/results.

Committee on the Global Financial System. 2011. “The Impact of Sovereign Credit Risk on Bank Funding Condition.” CGFS Papers 43, Bank for International Settlements, Basel, Switzer-land. http://www.bis.org/publ/cgfs43.htm.

Constâncio, V. 2013. “Establishment of the Single Supervisory Mechanism: The First Pillar of the Banking Union.” Speech at the 11th Annual European Financial Services Conference, Brus-sels, January 31. http://www.ecb.int/press/key/date/2013/html /sp130131.en.html.

Darracq Paries, Matthieu, Ester Faia, and Diego Rodriguez Palen-zuela. 2013. “Bank and Sovereign Debt Risk Connection.” SAFE Working Paper Series No. 7, Sustainable Architecture for Finance in Europe, Frankfurt, Germany. http://papers.ssrn .com/sol3/papers.cfm?abstract_id=2228494.

Das, Satiyajit. 2011. “The EBA Stress Tests— Not the Real Thing.” Eurointelligence, July 26.

Department of Finance— Government of Ireland, and Central Bank of Ireland. 2010. Letter of Intent, Memorandum of Economic and Financial Policies and Technical Memorandum of Understanding. Dublin, Ireland: Department of Finance— Government of Ire-land, December 3. http://www.imf.org/en/News/Arti cles/2015 /09/14/01/49/pr10496.

Dudley, W. 2011. “U.S. Experience with Bank Stress Tests.” Speech at the Group of 30 Plenary Meeting, Bern, Switzerland, May 28. http://www.newyorkfed.org/newsevents/speeches/2011 /dud110627.html.

European Banking Authority (EBA). 2011a. Capital Buffers for Ad-dressing Market Concerns over Sovereign Exposures: Methodological Note. London, UK: European Banking Authority, December 8.


https://www.eba.europa.eu/-/








http://business.time.com/2009/04/24/


http://www.guardian.co.uk/books/2008/sep/28/businessandfinance.philosophy

http://www.guardian.co.uk/books/2008/sep/28/businessandfinance.philosophy

https://www.newyorkfed.org/re

https://www.newyorkfed.org/re

http://search/staff_reports/sr409.html

http://www.imf.org/external/pubs/cat/longres.aspx?sk=22493.




http://www.imf.org/external/pp/longres.aspx?id=4702.

http://www.imf.org/external/pp/longres.aspx?id=4702.







https://www.centralbank.ie/publication/financial-measures-programme


http://www.imf.org/external/pubs/cat/longres.aspx?sk=24694.0





http://www.bis.org/publ/cgfs43.htm

http://www.ecb.int/press/key/date/2013/html/sp130131.en.html

http://www.ecb.int/press/key/date/2013/html/sp130131.en.html



http://www.imf.org/en/News/Arti

http://www.newyorkfed.org/newsevents/speeches/2011/dud110627.html

http://www.newyorkfed.org/newsevents/speeches/2011/dud110627.html








http://business.time.com/2009/04/24/



https://www.eba.europa.eu/-/cebs-s-statement-on-stress-testing-exercise

https://www.eba.europa.eu/-/cebs-s-statement-on-stress-testing-exercise

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Key-Aspects-of-Macroprudential-Policy-PP4803




Messenger.” Vox: CEPR’s Policy Portal. http://www.voxeu.org /article/defence-european-banking-authority.


Peristiani, Stavros, Donald P. Morgan, and Vanessa Savino. 2010. “The Information Value of the Stress Test and Bank Opacity.” FRBNY Staff Report 460, Federal Reserve Bank of New York, New York, NY. http://www.newyorkfed.org/research/staff_reports /sr460.html.

Pritsker, Matthew. 2010. “Informational Easing: Improving Credit Conditions through the Release of Information.” Economic Pol-icy Review. 16 (1): 77–87. http://data.newyorkfed.org/research /epr/2010.html.

Roland Berger. Stress Testing Spanish Banks. 2012. Madrid, Spain: Roland Berger Strategy Consultants, June 21. https://www .bde.es/bde/en/secciones/prensa/infointeres/reestructuracion /valoracionesind/.

Schuermann, Til. 2012. “Stress Testing Banks.” Wharton Finan-cial Institutions Center Working Paper 12-08, University of Pennsylvania, Philadelphia, PA. http://fic.wharton.upenn.edu /fic/papers/12/p1208.htm.

Schuermann, Til. 2013. “The Fed’s Stress Tests Add Risk to the Financial System.” Wall Street Journal, March 20. https://www .wsj.com/articles/SB10001424127887324532004578362543899602754.

Skidmore, Gregory. 2008. “Current Financial Crisis Is a Black Swan.” Seeking Alpha, October 6. http://seekingalpha.com/article/98669 - current- financial-crisis-is-a-black-swan.

Steinhauser, Gabriele. 2011. “EU: 2011 Bank Stress Tests to Be Tougher Than 2010.” The Washington Post, March 18. http://www.washingtonpost.com/wp-dyn/content/article/2011/03 /18/AR2011031800582.html?noredirect=on.

Tarullo, Daniel K. 2010. “Lessons from the Crisis Stress Tests.” Pre-sented at the Federal Reserve Board International Research Forum on Monetary Policy, Washington, DC, March 26. http://www.federalreserve.gov/newsevents/speech/tarullo20100326a .htm.

Wilson, Harry. 2011. “European Banking Authority Was Hobbled over Stress Tests.” The Telegraph, July17. http://www.telegraph .co.uk/finance/financialcrisis/8642492/ European- Banking - Authority-was-hobbled-over-stress-tests.html.

Wishart, Ian. 2011. “Stress Tests: Getting It Right This Time Around?” Politico- European Voice, February 9. https://www .politico.eu/article/ stress- tests- getting-it-right-this-time-around/.

Instruments— Initial Considerations: Report to the G- 20 Finance Ministers and Central Bank Governors. Washington, DC: Interna-tional Monetary Fund. http://www.fsb.org/2009/11/r_091107c/.

Irish Times. 2010. “Has Matthew Elderfield Restored the Credibil-ity of the Financial Regulator’s Office?” April 16, 2010, Irish Times, Dublin, Ireland.

Jenkins, P. 2011. “EBA Chief Stresses Rigour of New Bank Tests.” Fi-nancial Times, March 2011. https://www.ft.com/content/b5bf8ee8 -50c7-11e0-9227-00144feab49a#axzz2Oaias6ew.

Jobst, Andreas A., Li Lian Ong, and Christian Schmieder. 2013. “An IMF Framework for Macroprudential Bank Solvency Stress Testing: Application to S- 25 and Other G- 20 Country FSAPs.” IMF Working Paper 13/68, International Monetary Fund, Washington, DC. http://www.imf.org/external/pubs /cat/longres.aspx?sk=40390.0.

Langley, Paul. 2013. “Anticipating Uncertainty, Reviving Risk? On the Stress Testing of Finance in Crisis.” Economy and Society. 42 (1): 51–73. http://www.tandfonline.com/doi/abs/10.108 0/03085147 .2012.686719.

Lister, Tim, and Al Goodman. 2012. “Spanish Banks ‘Need 40 Bil-lion’ as Eurozone Ministers Prepare for Talks.” CNN, June 9. http://www.cnn.com/2012/06/09/world/europe/spain-imf.

Matsakh, Emil, Yigit Altintas, and Will Callender. 2010. “Stress Testing in Action: Lessons from the Financial Crisis: Undertak-ing a Robust Stress Test and Communicating Results in a Cred-ible Manner Will Enable Banks to Increase Investor Confidence.” Bank Accounting & Finance ( February- March): 18–27.

Ministry of Economy and Competitiveness, and Banco de España. 2012. Methodology and Results of the Independent Stress Tests.” Ma-drid, Spain: Banco de España, June 21. http://www.bde.es/bde /en/secciones/prensa/infointeres/reestructuracion/valoracionesind.

Mody, Ashoka, and Damiano Sandri. 2011. “The Eurozone Crisis: How Banks and Sovereigns Came to Be Joined at the Hip.” IMF Working Paper 11/269, International Monetary Fund, Washing-ton, DC. http://www.imf.org/external/pubs/cat/longres.aspx?sk =25363.

Oliver Wyman. Asset Quality Review and Bottom- up Stress Test Ex-ercise. 2012a. Madrid, Spain: Banco de España, September 28. https://www.bde.es/bde/en/secciones/prensa/infointeres /reestructuracion/valoracionesind/.

———. Bank of Spain Stress Testing Exercise. 2012b. Madrid, Spain: Banco de España, June 21. https://www.bde.es/bde/en /secciones/prensa/infointeres/reestructuracion/valoracionesind/.

Onado, Marco, and Andrea Resti. 2011. “European Banking Au-thority and the Capital of European Banks: Don’t Shoot the

© 2014 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).


http://www.voxeu.org/article/defence-european-banking-authority

http://www.voxeu.org/article/defence-european-banking-authority

http://www.mdpi.com/2227-7072/2/1/15

http://www.newyorkfed.org/research/staff_reports/sr460.html

http://www.newyorkfed.org/research/staff_reports/sr460.html

http://data.newyorkfed.org/research/epr/2010.html

http://data.newyorkfed.org/research/epr/2010.html

https://www.bde.es/bde/en/secciones/prensa/infointeres/reestructuracion/valoracionesind/



http://fic.wharton.upenn.edu/fic/papers/12/p1208.htm

http://fic.wharton.upenn.edu/fic/papers/12/p1208.htm

https://www.wsj.com/articles/SB1000142412788732453200457836254389

https://www.wsj.com/articles/SB1000142412788732453200457836254389

http://seekingalpha.com/article/98669-current-

http://seekingalpha.com/article/98669-current-

http://www.washingtonpost.com/wp-dyn/content/article/2011/03/18/AR2011031800582.html?noredirect=on



http://www.federalreserve.gov/newsevents/speech/tarullo20100326a.htm



http://www.telegraph.co.uk/finance/financialcrisis/8642492/

http://www.telegraph.co.uk/finance/financialcrisis/8642492/

http://Authority-was-hobbled-over-stress-tests.html

https://www.politico.eu/article/

https://www.politico.eu/article/

http://www.fsb.org/2009/11/r_091107c/

https://www.ft.com/content/b5bf8ee8-50c7-11e0-9227-00144feab49a#axzz2Oaias6ew

https://www.ft.com/content/b5bf8ee8-50c7-11e0-9227-00144feab49a#axzz2Oaias6ew



http://www.tandfonline.com/doi/abs/10.108

http://www.cnn.com/2012/06/09/world/europe/spain-imf

http://www.bde.es/bde/en/secciones/prensa/infointeres/reestructuracion/valoracionesind

http://www.bde.es/bde/en/secciones/prensa/infointeres/reestructuracion/valoracionesind

http://www.imf.org/external/pubs/cat/longres.aspx?sk=25363

http://www.imf.org/external/pubs/cat/longres.aspx?sk=25363

https://www.bde.es/bde/en/secciones/prensa/infointeres/reestructuracion/valoracionesind/.

https://www.bde.es/bde/en/secciones/prensa/infointeres/reestructuracion/valoracionesind/.



http://creativecommons.org/licenses/by/3.0/

http://www.tandfonline.com/doi/abs/10.108

PART III

Frameworks


CHAPTER 15

Macroprudential Bank Solvency Stress Testing in FSAPs for Systemically Important Financial Systems

ANDREAS A. JOBST • LI LIAN ONG • CHRISTIAN SCHMIEDER

The global financial crisis put bank stress tests squarely in the spotlight. Stress tests conducted in the lead- up to the crisis, including those by the IMF staff, were not always able to identify underlying risks and vulnerabilities and their magnitude. Since then, the IMF staff has developed more

robust stress testing methods and adopted a more coherent and consistent approach. This chapter articulates the solvency stress testing frame-work that is commonly applied in the IMF’s surveillance of member countries’ banking sectors, and discusses examples of its implementation of Financial Sector Assessment Programs (FSAPs) in 18 countries with systemically important financial systems or that are in the Group of Twenty, in the latter part of the global financial crisis. In doing so, the chapter offers guidance for readers seeking to develop their own stress testing frame-works and country authorities preparing for FSAPs. A detailed stress test matrix comparing the stress test parameters applied in each of these major country FSAPs is provided, together with stress test output templates.

This chapter is based on IMF Working Paper 13/68 (Jobst, Ong, and Schmieder 2013). The authors would like to thank Laura Kodres and country authori-ties for their helpful comments and IMF staff members who participate as stress testers in Financial Sector Assessment Programs (FSAPs) for their input.1 In November 2008, the G20 asked the IMF and the FSB to collaborate on a regular Early Warning Exercise (EWE). The EWE examines unlikely but

plausible, high- impact risks that would necessitate policy recommendations that could differ from those related to baseline projections presented in the World Economic Outlook, Global Financial Stability Report, and the Fiscal Monitor. The EWE does not attempt to predict crises. Rather, it seeks to identify (1) potential vulnerabilities that could precipitate systemic crises; and (2) suitable risk- mitigating policies, including those that would require interna-tional cooperation. It integrates macroeconomic and financial perspectives on systemic risks, drawing on a wide range of quantitative tools and broad- based consultations. (See also IMF 2018.)

quently transformed into the annual Comprehensive Capital Analysis and Review for large financial institutions. Ulti-mately, the crisis underscored that stress tests, irrespective of their level of sophistication or regularity of implementation, are not fail- safe, stand- alone diagnostic approaches, but need to be complemented with other tools, such as the Early Warning Exercise, which the IMF 2010d completes regu-larly together with the Financial Stability Board (FSB).1

Post mortems following the crisis show that the stress tests conducted by supervisory authorities, the IMF staff, and financial institutions themselves were not always able to identify imminent risks and exposures. As such, they fre-quently failed to provide sufficient early warning of potential

1. INTRODUCTIONThe global financial crisis has placed the spotlight squarely on the stress testing of financial institutions, notably that of banks. On one hand, the crisis revealed the shortcomings of stress tests as a tool for detecting important vulnerabilities during the lead- up period, which forestalled possible miti-gating actions being taken. On the other hand, the experi-ence highlighted the usefulness of credible stress tests in restoring market confidence in the financial system— by shedding light on the potential magnitude of capital shortfall— as demonstrated by the successful Supervisory Capital Assessment Program exercise undertaken by the US authorities in 2009 (Bernanke 2010), which was subse-


Macroprudential Bank Solvency Stress Testing in FSAPs for Systemically Important Financial Systems366

The communication of stress test results has also become an increasingly sensitive issue for the IMF’s membership. Both financial supervisors and financial institutions are struggling to balance the call for increased transparency with the need to avoid unduly alarming the markets and cre-ating self- fulfilling prophesies, especially during periods of stress.

In the decade- and- a- half since stress testing premiered in the IMF’s surveillance toolkit, the scope of stress tests has expanded to include both banking and nonbank institutions, with a strong focus on the former. However, the number of FSAP stress tests of the insurance and pension funds sectors, while having increased, still trails that of the banking sector significantly. The IMF staff has developed more robust stress testing methods and models, especially since the crisis. Based on the IMF’s rich and diverse practical experience with stress tests through more than a decade- and- a- half of FSAPs, the staff has proposed a set of “best practice” principles for macro- financial stress testing (Chapter 2 and IMF 2012a). The principles cover areas such as the scope of stress tests, transmission channels, risk types, valuation methodologies, risk sensitivity, and communication strategies.

At the IMF, bank stress testing is most advanced, given the systemic importance of the banking sector in practically all member countries. The focus has been on solvency risk, and work to continually develop a comprehensive and robust framework is ongoing. Separately, the development of li-quidity stress tests by the IMF staff, which is covered in Chapter 16, has also intensified in response to lessons learned from the crisis. This chapter complements Chapter 2 by pro-viding an operational perspective of those “best practice” principles for bank solvency stress testing applied by the IMF staff. Specifically, the chapter:

• Articulates the stress testing framework and demon-strates how best practice principles have been applied to key elements in the IMF’s surveillance of banking sec-tors in selected FSAPs. The sample group consists of 16 of the 25 countries with financial systems that had been identified at the time as being systemically important and were subject to mandatory assess-ments every five years (IMF 2010a, 2010b),3 plus two of the five Group of Twenty countries that are not among the systemic group (all hereafter “major countries”) (Table 15.1). These 18 jurisdictions par-ticipated in FSAPs during the latter part of the global financial crisis, between the 2010–13 IMF fiscal years, when the IMF staff was implementing signifi-cant changes to the stress testing framework. In fo-cusing on these countries during this period, the comparisons capture how IMF stress tests evolved with the aim of improving their robustness, trans-parency, and consistency.

vulnerabilities (Borio, Drehmann, and Tsatsaronis 2012). In some cases, the simulated shocks and resulting impacts were not sufficiently severe (often informed by historical stress events), reflecting the general reluctance to recognize the possible realization of extreme scenarios;2 in others, failure was attributable to the specification of the stress tests them-selves, including inadequate models to capture complex fi-nancial instruments, behavioral elements, or feedback effects. Elsewhere, inadequate data or weaknesses in scenario design, such as the exclusion or cursory treatment of certain types of risks and insufficient focus on spillover risks across different segments of the financial system— within a country as well as across borders— also contributed to the lack of ro-bustness of the stress tests.

At the IMF, stress testing has become a central aspect of the staff’s macroprudential surveillance of financial sys-tems. It is a key component of the Financial Sector Assess-ment Program (FSAP) and has evolved into an important part of the conjunctural analysis in the Global Financial Sta-bility Report; it is also applied in annual Article IV consulta-tion process and crisis program work. Also IMF member countries view stress testing as an essential aspect of supervision and financial stability analysis. In addition to microprudential (or supervisory) stress testing, some juris-dictions have established national macroprudential authori-ties, which engage in macroprudential stress testing. Countries are also increasingly requesting technical assis-tance on stress testing from the IMF as they seek to build or enhance their capacity in this area. These developments strengthen the case for a coherent and consistent approach to stress testing by the IMF staff in its engagement with the membership.

With more attention drawn to stress testing, exercises conducted by the IMF staff have come under considerable scrutiny. Consistency in the implementation of these stress tests is essential to enhanced disclosure and transparency, especially during volatile times. In this context, the comple-tion of stress tests as part of the financial stability assessment in FSAPs has resulted in several important observations.

The variety of approaches, models, scenarios, and assump-tions applied in the staff’s analyses has given rise to questions about the interpretation of the results and their consistent comparison across countries. The issue has been further com-plicated by the lack of generally accepted “best practice” prin-ciples (at least for some dimensions of stress tests) (see for example, Board of Governors of the Federal Reserve System, Federal Deposit Insurance Corporation, and Office of the Comptroller of the Currency 2012; BCBS 2018a) and evolv-ing practices. The IMF staff has developed prescriptive guide-lines in IMF- related stress testing exercises to ensure sufficient coverage and a modicum of uniformity, both within a finan-cial system and, at the very least, across “peer” countries.

2 Ultimately, authorities will have to define a specific level of risk toler-ance for their financial system (Hardy and Schmieder 2013).

3 In late 2013, the IMF increased the number of systemically important financial systems, from 25 in 2010 to 29 (IMF 2013).


Andreas A. Jobst, Li Lian Ong, and Christian Schmieder 367

of the various components and elements of the stress testing framework and their application in FSAPs. Section 4 concludes.

2. IMF STRESS TESTING IN CONTEXTStress testing is a forward- looking technique that attempts to measure the sensitivity of a portfolio, an institution, or even an entire financial system to events that have a very small probability of occurrence but a significant impact if they man-ifest. Scenario and/or sensitivity analysis are applied in “what if” exercises, which characterize what might happen if certain “ extreme- but- plausible” risks were to crystallize. In the decade and a half since the concept was first introduced, stress testing has been used by central banks, financial supervisors, and in-ternational organizations, such as the IMF, to identify vulner-abilities and incipient risks in the financial sector. Stress tests are commonly used for macroprudential, microprudential, and/or riskmanagement purposes (Figure 15.1).

Stress testing conducted by the IMF staff, as part of the institution’s surveillance mandate, is completed mostly in FSAPs and for macroprudential purposes (IMF and World Bank 2003; Moretti, Stolz, and Swinburne 2008). It is aimed at assessing system- wide resilience to shocks over the medium term, uncovering vulnerabilities to any rapid dete-rioration in the macroeconomic environment, and, more generally, identifying potential threats to overall financial stability. In this context, stress tests for both solvency and liquidity risks tend to incorporate very severe scenarios to assess the capacity of the financial system to withstand tail risks. The findings of these exercises typically do not require management action by financial institutions; rather, they are

• Presents the framework in a detailed cross- country stress testing matrix to compare the actual implementation across the major country FSAPs (Appendix 15.1). An abridged version of this stress test matrix for each country, which carries the technical reference “STeM,” is typically presented in the main FSAP report, the Financial System Stability Assessment (FSSA), to en-hance the transparency of each exercise.

• Aims to provide useful guidance for readers seeking to develop their own stress testing frameworks and for country authorities preparing for FSAPs. The chapter is illustrative in this regard in that it discusses the de-tailed setup of FSAP stress testing exercises.

Eight of the 18 countries in the sample published all the details of their respective FSAP stress tests. They comprise Australia, France, Germany, Japan, Spain, Sweden, the United Kingdom, and the United States. Of the remaining 10 coun-tries, all but one consented to the inclusion of all information on their respective FSAP stress tests, some of which was not contained in previously published reports.

There has been some success in standardizing FSAP stress tests across countries, and improvements continue to be made in this area. However, in some instances, expert judg-ment might render “one size fits all” approaches less relevant. Moreover, it is important to recognize that surveillance stress tests are not fail- safe, stand- alone diagnostic tools, al-though the value of well- designed exercises should not be underestimated. Their usefulness critically depends on the availability and quality of data.

This chapter is organized as follows. Section 2 puts into context the nature of the stress testing work conducted by the IMF staff. It is followed in Section 3 by detailed coverage

TABLE 15.1

S- 25 and Other G20 Countries: Status of FSAPs since Fiscal Year 2010Rank Jurisdiction Grouping Completed

FSAPs since FY 2010

Rank Jurisdiction Grouping Completed FSAPs since FY 2010

1 United Kingdom S- 25, G20, G7 FY 2011 16 Hong Kong SAR S- 25 —**2 Germany S- 25, G20, G7 FY 2011 17 Brazil S- 25, G20 FY 20123 United States S- 25, G20, G7 FY 2010 18 Russian Federation S- 25, G20 FY 20114 France S- 25, G20, G7 FY 2012 19 Korea S- 25, G20 —**5 Japan S- 25, G20, G7 FY 2012 20 Austria S- 25 FY 20136 Italy S- 25, G20, G7 FY 2013 21 Luxembourg S- 25 FY 20117 Netherlands S- 25 FY 2011 22 Sweden S- 25 FY 20118 Spain S- 25 FY 2012 23 Singapore S- 25 —**9 Canada S- 25, G20, G7 —** 24 Turkey S- 25, G20 FY 2011

10 Switzerland S- 25 —** 25 Mexico S- 25, G20 FY 201211 China S- 25, G20 FY 2010 Argentina G20 FY 201312 Belgium S- 25 FY 2013 European Union G20 FY 2013*13 Australia S- 25, G20 FY 2013 Indonesia G20 FY 201014 India S- 25, G20 FY 2012 Saudi Arabia G20 FY 201115 Ireland S- 25 — South Africa G20 —

Sources: IMF 2010a, 2010b, and 2013; and Monetary and Capital Markets Department, IMF.Note: Systemic- 25 (S-25) countries were ranked according to the size and interconnectedness of their financial systems. The IMF’s fiscal year (FY) runs from May 1 of the previous year to April 30 of the current year. G20 = Group of Twenty.*Stress tests were not conducted as part of the FSAP for the European Union.**FSAPs scheduled for completion in FY 2014.


Macroprudential Bank Solvency Stress Testing in FSA

Ps for Systemically Im

portant Financial Systems

368

Bank SolvencyStress Testing

Macroprudential

Surveillance CrisisManagement

Microprudential

Supervisory

Top Down(for example, CCAR)

Bottom Up(for example, CCAR)

Top Down(for example, SCAP,

CEBS/EBA,Spain, IMF crisis

programs)

Bottom Up(for example, SCAP,

CEBS/EBA)

Top Down(for example, FSAPs,GFSR, central bankfinancial stability

units)

Bottom Up(for example, FSAPs)

RiskManagement

Internal RiskManagement

Bottom Up(for example,

financialinstitutions’ own)

Main focus ofIMF stress tests

Source: Authors.Note: Top-down stress tests are completed based on either individual bank data (which are then aggregated) or aggregated portfolios; bottom- up stress tests are conducted by individual institutions using their own internal risk models and data. CCAR = Comprehensive Capital Analysis and Review; CEBS = Committee of European Banking Supervisors; EBA = European Banking Authority; FSAPs = Financial Sector Assessment Programs; SCAP = Supervisory Capital Assessment Program.

Figure 15.1 Solvency Stress Testing Applications



determine the condition of the banking sector as an input to the design of a potential program.

Separately, financial institutions regularly carry out stress tests for riskmanagement purposes. In these internal exer-cises, financial institutions develop and implement their own stress testing programs, which assess their ability to meet capital and liquidity requirements under stressed con-ditions. The IMF staff sometimes relies on banks’ stress test-ing infrastructure for the FSAP bottom- up stress tests (see Section 3). In some countries, supervisors issued guidance on stress testing to the financial institutions under their su-pervision before FSAPs took place (for example, the United Kingdom and the United States) (Board of Governors of the Federal Reserve System 2012c; Financial Services Authority 2009). However, this practice is not yet widely implemented, including in some of the world’s largest financial systems. The Basel Committee for Banking Supervision (BCBS) has also issued guidelines for stress testing by individual banks (BCBS 2009, 2017), followed up by a peer review of supervisory authorities’ implementation of those principles (BCBS 2012a).

3. A FRAMEWORK FOR BANK SOLVENCY STRESS TESTINGThe objective of the bank solvency stress tests conducted by the IMF staff is to assess the soundness of banking sectors under adverse macroeconomic conditions. Tests are designed to determine banks’ resilience to the adverse impact of severe macro- financial stress over the short and medium terms (rel-ative to a predefined baseline scenario). The aim is to identify the sector’s vulnerabilities and its capacity to absorb shocks.

Within the framework, the development of plausible and coherent tests requires a thorough understanding of the fi-nancial system in question and its institutions, which in-cludes structural characteristics, such as differences in banks’ business models, their role in the domestic financial sector, and, increasingly, cross- border linkages. While the identifi-cation of transmission channels of stress impacting financial intermediation in smaller countries tends to be straightfor-ward, more complex banks in larger economies and financial centers may create conceptual challenges for stress testing.

Up until the latter part of the global financial crisis, in 2013, the IMF had conducted assessments of about 140 ad-vanced and developing countries. Of these, solvency stress tests had been conducted in practically all instances. Thus, the solvency stress tests in FSAPs must necessarily be adapt-able to diverse financial systems, with consistent applications of assumptions and models but sufficient flexibility to ac-commodate vastly different circumstances (for example, normal or crisis times), systems (for example, sophisticated or basic), and regulatory regimes (for example, Basel I or Basel II/III) as well as be sensitive to when and how the out-comes are presented and communicated (Table 15.2). Further, the FSAP stress tests necessarily require trade- offs

used to inform policy discussions with country authorities about the frameworks in place to deal with systemic shocks.

The cooperation with country authorities has a crucial impact on the robustness and credibility of IMF stress tests (see Section 3). According to Article VIII of the Articles of Agreement of the IMF, member countries are under no obli-gation to disclose information about individuals or corpora-tions. This means that the IMF cannot compel country authorities to provide the necessary confidential bank- by- bank data for the stress tests. In some cases, authorities have refused to share any prudential information, and the IMF staff has had to rely solely on publicly available data, which reduces the specificity of the results; in others, authorities have only consented to running the tests themselves, based on some agreed- upon parameters, and sharing the aggre-gated results. The recourse for the IMF staff is to ensure that the transparency of the process— or any limitations thereof— is clearly documented in the official documents.

The IMF’s objectives may be contrasted with the stress tests undertaken by supervisory authorities, usually for mi-croprudential purposes (Fell 2006). Such exercises are nor-mally embedded in the oversight process wherein supervisors would run stress tests involving individual institutions on a periodic basis to assess their financial soundness under ad-verse economic conditions, such as in the case of the annual US Comprehensive Capital Analysis and Review (Board of Governors of the Federal Reserve System 2012a, 2012b), which is completed together with a liquidity stress test (Com-prehensive Liquidity Analysis and Review). Supervisory stress tests can be independent of an institution’s systemic rele-vance, where “failure” would typically require some form of management action, including recapitalization.

The global financial crisis brought about a new concept of stress testing, that is, one with a crisis management objec-tive, which the IMF staff refers to as “crisis stress testing” (see Chapter 14). Largely macroprudential, as the aim is to restore and sustain market confidence in the financial sys-tem, it can also be considered microprudential in that it ex-amines the soundness of individual financial institutions, and “failure” would typically require recapitalization— or even resolution. Compared to the medium- term risk horizon of surveillance stress tests, crisis stress tests tend to have a near- term focus. In the United States, system- wide (sol-vency) stress testing of banks was used by the authorities in 2009 for crisis management purposes, through the Supervi-sory Capital Assessment Program exercise (Board of Gover-nors of the Federal Reserve System 2009), the predecessor of the Comprehensive Capital Analysis and Review; the EU authorities also made a similar effort through the region- wide stress testing exercise conducted by the Committee of European Banking Supervisors in 2009 and 2010 and then by its successor, the European Banking Authority in 2011 (Committee of European Banking Supervisors 2010; EBA 2010, 2011a, 2011b), as did Ireland (Central Bank of Ireland 2011) and Spain (Banco de España 2012). IMF teams work-ing on crisis countries may sometimes run stress tests to



testing approach(es); (2) the coverage in terms of the institu-tions, their market shares, and the sources of their earnings and exposures; and (3) the granularity and timeliness of rel-evant data (and their reliability). In this regard, stress tests conducted by the IMF staff for financial surveillance pur-poses are typically undertaken in close collaboration with supervisory authorities. In many instances, the staff is given access to the necessary detailed supervisory data during FSAPs (on agreement of strict confidentiality); data quality

among the scope, scenario design, and methodologies ap-plied in the context of staff and authorities’ resources and time constraints.

Scope

The scope of a stress testing exercise needs to be sufficiently comprehensive to capture the main characteristics of a par-ticular financial system. Key considerations are: (1) the stress

TABLE 15.2

A Framework for Macroprudential Bank Solvency Stress TestingFramework/Components Key Elements Illustrative Example1. Scope

Approach Bottom- up (BU) By individual banks Top- down (TD) By authorities; by IMF

Coverage Institutions Number of banksMarket share Percentage of banking sector assets

Data Source Banks’ own, supervisory, and public dataCutoff date End of last fiscal yearReporting basis Unconsolidated banking groups, domestic

businesses only2. Scenario design

Risk horizon MultiperiodInstantaneous

1–5 years

Scenarios Baseline IMF World Economic Outlook projectionsGrowth shocks Double- dip recession and protracted slow growth

Risks Key risk(s) Credit risk, market riskOther risks covered in

scenario analysisSovereign risk, funding risk, exchange rate risk

Other tests/risks Sensitivity analysis of credit and market risks; network analysis of spillover risk

Factors that management controls

Balance sheet growth Consistent with nominal GDP growth

Credit growth Based on satellite modelDividend payout rule Historical payout ratioOther business strategy

considerationsNo asset disposal allowed

Other assumptions Taxes Uniform (local corporate income) tax rate

3. Regulatory capital standardsCapital definition Domestic

InternationalLocal regulatory requirementsBasel III transition

Capital adequacy Metrics Amount of recapitalization required (in domestic currency); total capital, Tier 1 and core/common equity Tier 1

Hurdle rate(s) In line with Basel III transition scheduleChanges in risk- weighted

assets Risk- weighted assets calculated using Basel II formula

4. MethodologyStress test model Accounting- based Balance sheet approach (for example, Schmieder and others 2011)

Market price- based Systemic Contingent Claims Analysis (Jobst and Gray 2013)

Modeling of macro- financial linkages

Satellite models Econometric models for credit losses, income, credit growth

5. CommunicationPresentation of output Template(s) Standardized output template for individual BU results provided to banks and

authoritiesPublication Medium Results published in FSSA; Technical Note published

Source: Authors.Note: FSSA = Financial System Stability Assessment.



Australia, China, France, India, Indonesia, Japan, Mexico, Russia, Turkey, and the United Kingdom). TD tests were either conducted by the IMF team only (for example, Aus-tralia, India, Indonesia, Netherlands, Saudi Arabia, Sweden, and Turkey) or by the authorities only (for example, Japan, Luxembourg, and Russia), or in some cases, separately by both, using different methods (for example, China, France, Mexico, and the United Kingdom).

The solvency stress testing of the banking sector in the 2011 FSAP Update for the United Kingdom exemplifies the effective collaboration among country authorities, the IMF, and individual financial institutions in all aspects of the exercise (IMF 2011a), which involved both BU and TD sol-vency stress tests (together with TD liquidity risk stress tests). BU stress tests were run by the seven major UK banks, in close coordination with the FSAP team and the then Fi-nancial Services Authority (Figure 15.2). At the same time, TD tests were separately performed by the Bank of England using its Risk Assessment Model for Systemic Institutions and by the FSAP team using the Systemic Contingent Claims Analysis model, applying macroeconomic forecasts and projections from the IMF and FSAP, respectively, and satellite model outputs from the Bank of England.

Coverage

The coverage is crucial for the usefulness, and, thus, credibil-ity of the stress test exercise. Ideally, surveillance stress test-ing for macroprudential purposes should include all institutions, if data availability and resources permit. Realis-tically, all systemically important institutions, as well as second- tier banks that are potentially systemic depending on circumstances, should be covered. Smaller banks that may be considered at risk could also be included.

FSAPs typically focus on stress testing the major com-mercial banks in their respective jurisdictions, with coverage usually determined in collaboration with the authorities. Where resource constraints dictate that only a small sample of banks can be considered, especially in the case of BU stress tests, the usual practice is to focus on the systemically important institutions. The market share of the banks in-cluded in the 18 major country stress testing exercises be-tween FY2010/13 was at least 60 percent of the total assets of the sector; coverage was 100 percent in six of them (Bra-zil, India, Indonesia, Japan, Luxembourg, and Russia).

The identification of systemically important domestic banks is still not clear- cut. While some banks are of obvious systemic importance in their own respective countries and their selection for stress tests is indisputable, the difficulty has been in identifying those that are systemic at the mar-gins, for example, some of the smaller institutions that may have the potential to become systemic depending on the environment at a particular point in time (IMF, Bank for International Settlements, and Financial Stability Board 2009). Thus, the definition of what constitutes a systemic

is further enhanced when individual financial institutions participate in the exercise.

Approach

Surveillance stress testing of banks’ solvency risk in the con-text of FSAPs is usually based on a “ top- down” (TD) ap-proach, which is sometimes combined with a “ bottom- up” (BU) exercise:

• TD tests are carried out by the IMF staff, by the au-thorities, or by both, typically in close collaboration. In these exercises, tests are either conducted using the data of individual banks (which are then aggre-gated), or on an aggregated group of banks to ana-lyze the impact of predefined, system- wide shocks. A common macro- financial environment is assumed, and a standardized set of behavioral assumptions is applied to all institutions. TD stress tests may be used as a stand-alone analysis or to complement the BU exercise, if one is conducted.

• The BU approach is used by FSAP teams if authori-ties are supportive of having individual institutions conduct their own stress tests (and banks are pre-pared and have sufficient capacity to do so). Individ-ual institutions use their own data and internal risk models. As with the TD approach, common macro-economic shocks and selected standardized assump-tions are prescribed by the IMF staff to isolate the impact of shocks on banks’ financial soundness to identify specific vulnerabilities.

The IMF staff advocates conducting both BU and TD stress tests, as much as possible, to enrich the surveillance analysis in FSAPs. Each approach has its strengths and weaknesses and is considered complementary for cross- validation purposes, rather than being considered a substi-tute for the other. The process of reconciling the BU and TD results is usually an important learning process in itself, with any divergence in the results from the two approaches usually traced to differences in either the model design, the scope of the stress testing exercise (including the type of underlying data used), behavioral assumptions, and/or modeling of sensitivities. For instance, bank- specific as-sumptions and the application of internal models based on more granular data can lead to differences in the projection of profits and losses— and consequently the impact on the capital ratios— for individual banks under the various scenarios.

The decision as to whether BU stress tests are conducted to complement TD tests, or whether TD stress tests are per-formed by country authorities or by the IMF staff, or jointly, is mostly made on an ad hoc, country- by- country basis, de-pending on data and resource availability as well as the re-ceptiveness and degree of involvement by authorities. Around half of the FSAPs for major countries between FY2010/13 comprised both BU and TD tests (for example,



mine the appropriate cutoff date. Supervisors typically make available to IMF teams the relevant data from prudential re-porting and reviews, which are usually supplemented by pub-licly available information. Supervisory data were provided in almost all 18 major country FSAPs covered in this chapter; only in one instance was the staff wholly dependent on public information for the stress testing exercise. If there is no access to supervisory data, the use of publicly available data on indi-vidual banks may be less granular and timely.

There has been little standardization across FSAPs re-garding the reporting level of bank data. While about half of the FSAPs in the sample used consolidated banking group data for the stress tests (for example, Australia, Brazil, China, France, Japan, Netherlands, Sweden, the United Kingdom, and the United States), the rest utilized uncon-solidated, legal entity data (for example, Germany, India, Luxembourg, Mexico, Russia, Spain, and Turkey).

FSAPs typically focus on the domestic banking sector, which suggests that the data of banks’ local businesses should be utilized on a local- consolidated basis. Such data would avoid double counting local business operations. The use of consolidated level data would prevent the examina-tion of ring- fencing of subsidiary profits, capital, and liquid-ity by host countries, which may be important for large international groups (Cerutti and Schmieder 2012). That said, the decision as to which type of data to use may some-times be moot, as it could be constrained by the type of data that are collected for supervisory purposes.

bank remains somehow ad hoc in IMF- related stress testing exercises, and a more structured approach is desirable. The BCBS methodology for identifying global systemically important banks, which has been reviewed recently, has fa-cilitated this process (BCBS 2011, 2018b; Financial Stability Board 2011). The guidelines on the implementation of su-pervisory measures for domestic systemically important banks and the policy recommendations by the Financial Sta-bility Board (Financial Stability Board 2012) for their identification represent another positive step in this direc-tion (BCBS 2012d), whereby many jurisdictions have identi-fied these banks in their jurisdictions.4

Data

The credibility of stress test results depends on the availability and sufficiency of timely and reliable data. The quantity and quality of data not only determine the scope and risk cover-age of the stress tests but also the type of models that can be applied (Howard 2008). As much as possible, FSAP stress tests utilize the latest audited and/or supervisory data along-side the latest macroeconomic projections, all of which deter-

4 In 2012, the Basel Committee developed a set of principles on the as-sessment methodology and the higher loss absorbency requirement for domestic systemically important banks. The framework takes a comple-mentary perspective to the global systemically important bank frame-work by focusing on the impact that the distress or failure of banks will have on the domestic economy.

Source: IMF 2011d.Note: CCA = contingent claims analysis; FSA = Financial Services Authority; FSAP = Financial Sector Assessment Program; LCR = liquidity coverage ratio; NSFR = net stable funding ratio; RAMSI = Risk Assessment Model of Systemic Institutions.

Figure 15.2 Example of IMF Stress Testing Exercise: UK FSAP Update

MACROPRUDENTIAL SURVEILLANCE STRESS TESTS

Solvency Liquidity

Bottom-up bybanks

Firms completeown stress test

according toIMF-developedguidelines, incoordinationwith the FSA

Top-down byauthorities

Bank of EnglandRAMSI

Six largest banks+ largest

building societyCoverage

Specification

TypeTop-down byFSAP team

IMF Systemic CCA

Top-down byauthorities

The FSA completes IMF-designedstress tests (five- and 30-day impliedcash flow; LCR and NSFR) using data

from Liquidity Reporting Profile

Five largest banks Six largest banks+ largest

building society

Six largest banks + five largestbuilding societies/mutuals + fivelargest foreign investment banks

Forecasts of income, operatingexpenses and credit losses



line scenario. In FSAPs, the IMF’s World Economic Outlook projections are typically used as the baseline for stress tests. Stress scenarios are then defined based on either (1) histori-cal simulation (by defining the scope and severity of the sce-nario based on previous stress episodes and crisis periods), or (2) hypothetical scenarios (which have not yet happened but are particularly relevant given specific vulnerabilities in banks’ portfolios). In both cases, ad hoc expert judgment of-ten influences the scenario design. The stress scenarios are then applied consistently across banks within the same sector.

The availability of data and the modeling capabilities govern how the appropriate stress scenarios are constructed for FSAP solvency tests. Scenarios either (1) reflect a hypo-thetical state of risk parameters under stress affecting sol-vency conditions (“direct approach”), which is often used in the case of ad hoc scenarios or historical simulation, or (2) are based on adverse macroeconomic scenarios, which need to be translated into financial stress parameters (“indi-rect approach”). The latter approach consists of the follow-ing elements:

• Estimating relevant macroeconomic and financial vari-ables conditional upon the chosen scenario. Common methods for predicting economic and financial vari-ables conditional upon certain macroeconomic condi-tions include: (1) structural econometric models, (2) vector autoregressive methods, and (3) pure statistical approaches (Foglia 2009). As a general rule, these macro- financial linkages would need to be clearly doc-umented and back- tested.

• Converting these variables into financial risk parameters via various types of “satellite” (or auxiliary) models. This step links different macro- financial shocks, reflected in macroeconomic variables, to the main determi-nants of bank solvency, that is, preimpairment profit, impairments, and risk- weighted assets (RWAs), since macroeconomic models do not usually include finan-cial balance sheet variables (and credit aggregates in particular). Common explanatory variables include:– Macroeconomic variables, such as economic

growth, unemployment, short- and long- term in-terest rates, inflation, and exchange rates;

– Sectoral (asset price) indicators, such as residential and commercial real estate prices, commodity prices, and equity market conditions (Figure 15.3); and

– Microlevel data, such as bank- specific credit growth (for example, deleveraging under severe stress con-ditions), which could also be modeled as a macro-economic variable, operational/financial leverage, and funding gaps.

FSAPs attempt to introduce consistently severe macro-economic shocks in the specification of scenarios in solvency stress tests. The aim is to facilitate the identification of other factors that drive differences across institutions and to facili-tate comparisons between peer countries. Shocks to

There is an increasing use of forward- looking market data and other variables to complement accounting infor-mation, especially for data- rich advanced economies. Mar-ket data reflect important investor perceptions of how risks affect the actual valuation of reported book values. They can also be used as a benchmark for the calibration of banks’ own credit risk models under internal- ratings- based (IRB) approaches— and for cross- validating the quantifica-tion of other, difficult- to- model risks, namely, market and operational risks.

The use and interpretation of the data require caution. FSAPs do not conduct audits of banks’ accounts and cannot corroborate the quality of the reported data and valuations used in stress tests. Thus, quantitative approaches benefit greatly from discretionary expert judgment. In instances where the staff may be concerned about the effects of issues such as loan misclassifications and/or lender forbearance on the accuracy of the data, caveats are often explicitly noted (for example, in the Spain and the United Kingdom exercises).

Scenario Design

Risk Horizon

For surveillance purposes, the choice of a risk horizon should be consistent with the desired implications for the related policy discussions. Covering a longer time period offers sev-eral benefits, in particular: (1) major macro- financial shocks typically have a lasting impact over several years, especially in the case of credit risk; and (2) evolving regulatory reforms are likely to be protracted and take several years to implement (for example, the implementation of the postcrisis regulatory reforms in Basel III). A longer risk horizon entails greater un-certainty, but surveillance stress testing is not a forecasting exercise; rather, the exercise should adequately capture any medium- term impact. In contrast, sensitivity tests are usually applied to assess instantaneous shocks.

It is important to balance the consistency of the risk horizon across countries with the usefulness of the findings for individual country circumstances. As in other aspects of stress testing, some expert judgment is involved— major country FSAPs typically apply a five- year risk horizon, but exceptions may be made in cases where the staff is of the view that the application of a longer sample period may be uncon-structive. As an example, the FSAP stress test for Spain dur-ing the global financial crisis applied a two- year risk horizon to accommodate the rapidly changing banking sector due to ongoing restructuring efforts (IMF 2012b). In most emerg-ing market economies with less mature banking sectors (for example, China, Indonesia, Mexico, and Turkey), risk hori-zons of between one and three years were used.

Stress Scenarios

Stress tests are based on scenario shocks and/or sensitivity analysis. In scenario tests, a baseline scenario is first estab-lished; postshock assessments are made relative to the base-



a separate stress (for example, Australia, Brazil, China, Ger-many, Japan, Sweden, Turkey, the United Kingdom, and the United States).

While the standardization of the scale of shocks has become a general rule of thumb for FSAPs (Hardy and Schmieder 2013), a certain flexibility in the scenario design remains key. The prevailing macroeconomic environment and market conditions as well as the main risks to financial stability determine the configuration of the most appropri-

economic growth are defined in terms of historical volatility, usually one (mildly adverse) and/or two (severely adverse) standard deviations from the long- term average. Among the sample countries covered in this chapter, the four- standard- deviation shock imposed on the Australian banking sector was estimated over a 50-year period, whereas the two- standard- deviation shocks applied to several EU countries were calculated over 30-year periods. In about half the exer-cises, a prolonged slow growth scenario was also included as

Source: IMF 2011d.Note: BoE fan charts are based on BoE, rather than WEO projections. BoE = Bank of England: CPI = consumer price index; DCLG = Department of Communities and Local Government; FSA = Financiall Services Authority; FTSE = Financial Times Stock Exchange; FSAP = Financial Sector Assessment Program; IPD = Investment Property Databank; WEO = World Economic Outlook.

Figure 15.3 Example of Macroscenarios for Stress Testing: UK FSAP Update

IMF mild double dip recession (DD mild) scenario = European Banking Authority (EBA) 2011 adverse scenario

IMF prolonged slow growth (SG) scenarioIMF severe double dip recession (DD severe) = FSA 2011 anchor scenario

IMF projected baseline scenario

–1–3–5–7

97531

5

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

5

13

11

9

7

2. Unemployment (In percent)

WEO baseline

(SG)DD mild

DD severe

–1

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

5

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

5

765

3

1

4

2

0

3. CPI (In percent change, year over year)

Baseline

SGDD mild

DD severe

–5–20–35

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

5

7055402510

4. FTSE All Share (In percent, year over year)

–5–15–25–35

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

5

3525155

6. IPD Commercial Property Price Index (In percent change, year over year)

BoE fan chart (Feb. 2011),90 percent density

BaselineSG

DD mild

BoE fan chart (Feb. 2011),90 percent density

Baseline

SG

DD severe

DD mild

Baseline

SGDD mild

DD severe

3.0

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

55.5

4.5

5.0

4.0

3.5

8. 10-Year Treasury Note Rate (In percent)

–5–10–15

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

5

201510

50

5. DCLG House Price Inex (In percent change, year over year)

Baseline SG

DD mild

DD severe

0

Q1:2

008

Q1:1

5Q3

:14

Q1:1

4Q3

:13

Q1:1

3Q3

:12

Q1:1

2Q3

:11

Q1:1

1Q3

:10

Q1:1

0Q3

:09

Q1:0

9Q3

:08

Q3:1

5

876

4

2

5

3

1

7. Three-Month Libor Rate (In percent)

BaselineSG

DD mild

DD severe

Baseline

SG

DD severe

DD mild

1. Real GDP (In percent change, year over year)



of risk factors in FSAP stress tests has evolved and expanded over time, with significant enhancements in the technical analysis and specification of risks in the wake of the global financial crisis. FSAPs attempt to cover all relevant macro- financial risks affecting the performance and valuation of banks and the financial system at large. Prior to the global financial crisis, these tests focused largely on credit and mar-ket risks (for example, interest rates, exchange rates, and credit spreads as well as equity and commodity prices). While these risks remain the mainstay of FSAP solvency stress tests, additional facets of risk affecting different types of exposures have been included:

• Exposures to sovereign and other previously low- default assets: Prior to the global financial crisis, exposures to sovereign debt did not figure prominently in stress tests, if at all. They were considered “risk free” and were typically assigned the lowest (often zero per-cent) risk weights for the calculation of regulatory capital requirements under the Basel framework. However, FSAPs have acknowledged rising sovereign risks by estimating the potential valuation losses of such exposures (and the commensurate impact on unexpected losses reflected in higher risk weights). The valuation loss can be derived from the negative impact of higher sovereign credit risk on the price of sovereign exposures, which would result in a corre-sponding valuation haircut (Chapter 10). The same approach can be applied to other previously low- default portfolios, such as holdings of bank debt (which are indirectly affected by higher sovereign risk). Shocks to sovereign exposures were incorpo-rated into the FSAP stress tests for all larger EU Member States and Japan; the same treatment was also applied to portfolios of bank debt in cases where domestic banks were highly exposed to their local government.

• Banking and trading books: For securities, stress tests had previously considered shocks to trading books only, largely because longer risk horizons were not covered. However, during the global financial crisis, many institutions moved their securities to their banking books, supported, in some cases, by regula-tory forbearance. This change in the accounting treatment underscored the need for stress tests to cover also (largely unrealized) valuation losses of se-curities in the two components of the banking book: (1) the available- for- sale portfolio, where losses are absorbed by reserves as part of shareholders’ equity (unlike mark- to- market losses in the trading book, which are reflected in net income), and (2) the hold- to- maturity portfolio, where lower market values (and higher downgrade risk) require higher provi-sions (which weigh on net income). However, not all country authorities are receptive to a comprehensive application of shocks to banks’ all securities hold-ings. In the FSAP stress tests for France, Japan, the

ate and credible tail- shock scenario(s). For example, the issue of overheating was a key risk for Turkey at the time of its 2011 FSAP and was therefore incorporated into the design of the stress scenario. For Spain, the one- standard- deviation shock applied during the global financial crisis included a revised baseline, which had already incorporated the rapidly deteriorating economic outlook and a fiscal adjustment (and, thus, implied a severity that was higher than that of a customary two- standard- deviation shock calibrated over a long- term average growth rate).

Nonetheless, spillover effects have remained largely un-addressed in the scenario design. For many years, the IMF’s stress tests did not consistently and comprehensively quan-tify the possible impact of scenarios on the macroeconomic conditions of other countries where stressed banks form part of the host banking sector. In such cases, the IMF staff often relied on the banks themselves to estimate the correspond-ing scenarios in relevant countries in BU exercises, poten-tially giving rise to inconsistent projections, and creating biased results of banks’ financial performance, possibly for the same countries. However, in the meantime, this aspect of IMF stress testing has been improved with the develop-ment of the Global Macro- Financial Model (Chapter 4).

FSAP stress scenarios emphasize the importance of tail risks. The tests are aimed at identifying the vulnerabilities of a country’s financial system and the ability of its supervisory and crisismanagement frameworks to deal with the realiza-tion of extreme but plausible risks. In the 2011 UK FSAP, for instance, capital losses were estimated for a 0.1 percent probability event (IMF 2011d, 2011c)—the UK Financial Services Authority 2011 had ascribed a 2 percent probability to a two- standard- deviation shock to growth materializing, and the IMF’s model subsequently calculated capital losses at the 95th percentile of this scenario (that is, falling into the 5 percent tail of the loss distribution), which would have a probability of 0.05 × 0.02 = 0.001 percent. That said, it is sometimes difficult to convince national authorities of the importance of running extreme tail scenarios. A useful way forward may be to also run reverse stress tests, that is, stress tests that aim to determine scenarios that would cause a bank to become insolvent.

Separately, sensitivity tests provide useful information on the immediate impact of individual shocks. They are usually applied if there is little or no data (and/or data quality is in-sufficient), or to complement the scenario analyses con-ducted on more complex financial systems. Several risk factors could also be combined to determine the impact of concurrent multiple shocks to a system. Sensitivity analyses were conducted in most major country FSAP stress testing exercises on various market risk factors.

Risk Factors

The selection of main risk drivers, and the way they are inte-grated (or not), has significant bearing on the interpretation and potential implications of stress test results. The coverage



risk parameters provide a more realistic assessment of ex-pected credit risk, especially during stressed periods. An-other key challenge in FSAPs is to ensure the availability of these parameters for all (or most) sample banks, and if neces-sary, develop proxy metrics, such as loan loss provisions (Schmieder, Puhr, and Hasan 2011).

Factors that Management Controls

Surveillance stress tests often include common assumptions about strategic decisions and behavioral adjustments of banks during times of stress. These assumptions (on factors that management controls) ensure that stress test findings can be analyzed in a consistent and comparable manner. In FSAPs, common assumptions are especially pertinent for BU stress tests that rely on banks’ internal models— at the expense of methodological flexibility and realism. Assump-tions adopted in FSAPs are also typically (and appropriately) on the conservative side. The main behavioral variables include:

• Balance sheet growth: This assumption determines the trend growth in core items on the assets and lia-bilities sides of banks’ balance sheets. FSAP stress tests typically assume that the balance sheet is either (1) constant (that is, growing with nominal GDP or some predefined rule); or (2) static (that is, not grow-ing at all, possibly in combination with a constant credit portfolio). Indeed, the major country FSAPs in the sample were split almost evenly on the adop-tion of either assumption.

• Credit growth: Assumptions about credit growth are usually based on either models (for example, Brazil, Spain, and Sweden) or descriptive empirical evidence (for example, Turkey), and, in many cases, also in-volve expert judgment. Banks under stress are likely to reduce lending in line with a slowdown or reversal in balance sheet growth, usually consistent with changes in nominal GDP.

• Dividend payout: Dividends are generally assumed to be paid only by banks that satisfy all three measures of capital adequacy, as relevant (that is, total capital, Tier 1, and core/common equity Tier 1) after mak-ing adequate provisions for asset impairments and transfers of profits to statutory reserves, which banks must keep on hand to meet their obligations to de-positors. In most of the major country FSAPs cov-ered in this chapter, it was assumed that banks would suspend dividends under stress. For the others, as-sumptions included payouts based on Basel III capi-tal conservation standards (for example, Sweden) or on historical ratios (for example, Brazil, France, and Japan).

• Strategic changes and asset disposal: FSAP stress tests typically do not consider changes to business opera-tions that require managerial involvement, such as plans to increase operational efficiencies. Moreover,

Netherlands, Sweden, and the United Kingdom dur-ing the global financial crisis, valuation haircuts were applied to both portfolios (excluding exposures to “AAA”-rated and/or own sovereigns in the hold- to- maturity portfolio), but only to the available- for- sale portfolio in the case of Russia and Spain.

• Funding costs: The global financial crisis underscored the importance of assessing the impact of rising funding costs on bank solvency (as part of the simu-lation of income under stress more generally). Fund-ing costs change disproportionately to changes in solvency conditions, rising sharply as a bank’s capital adequacy worsens (especially for banks with sizeable portions of wholesale funding, which is more risk sensitive than a large deposit base). Stress test calcu-lations link net funding costs (simulating the impact on both assets and liabilities) to income, possibly linking both solvency and liquidity exercises. The explicit treatment of funding costs in FSAP solvency stress tests was incipient during the global financial crisis (for example, France, Germany, Sweden, and the United Kingdom) but has since become an im-portant aspect of the exercise.

• Off- balance- sheet items: The realization of contingent liabilities from explicit and implicit guarantees of in-vestment vehicles (and the emergence of contingent assets related to related party lending during the global financial crisis) resulted in the sudden realiza-tion of large losses (and considerable cash outflows). Thus, incorporating off- balance- sheet positions that could give rise to such contingent liabilities (such as guarantees, commitments, and derivatives) is impor-tant to adequately capture the impact of extreme stress on all relevant exposures. However, such data are not as readily available, especially from public sources.

• Cross- border exposures: Prior to the crisis, credit risk tests focused largely on banks’ exposures to domestic firms and households without consideration of for-eign exposures (through branches and subsidiaries). Since then, FSAPs have incorporated spillover risks in the form of network (for example, Australia, France, Japan, and Spain) and ring- fencing (for example, Spain) analyses as separate modules in the TD approach. In some BU assessments, interna-tional banks have also considered shocks to the countries in which they are active (for example, the United Kingdom).

The estimation of impairment losses based on credit risk parameters— probability of default (PD) and loss given de-fault (LGD)—took on significant importance during the crisis. Differences in banks’ respective business models and/or specific risks tend to explain differences in the character-istics of nonaccrual events and impairment losses. The calibration approach of credit risk parameters— through- the- cycle or point- in- time PD (and LGD)—has a significant impact on the estimated net income of banks. Point- in- time



Sweden); in a couple of cases, their own national transitional schedules were applied (for example, Brazil and Japan);

• Used benchmark parameters from the BCBS’s Sixth Quantitative Impact Study (BCBS 2010c)—a com-prehensive study to evaluate the impact of the initial Basel III package agreed to in 2010—to simulate the likely impact of regulatory reforms on bank solvency (for example, Germany and the United Kingdom, where a separate and additional transitioning ar-rangement was also included for the BU exercise in the form of the interim capital regime); or

• Applied a separate local regulatory capital definition (for example, Mexico).

Also capital metrics (and appropriate hurdle rates) tend to vary across countries. For countries with solvency regimes based on the Basel II capital definition, FSAP stress tests ap-ply total regulatory capital to determine the hurdle rate. In cases when the Basel III (or a national modified version) applies, the metrics usually comprised the fully phased- in total capital, Tier 1 capital, and common Tier 1 capital (as of 2019 and 2022, respectively), along with the associated Basel III hurdle rates (Table 15.3) or (higher) national re-quirements, respectively. In some instance, the hurdles rates have included the capital conservation buffer (for example, France and Japan), while the loss absorbency requirement for global systemically important banks was applied rarely (for example, France). In a few cases, hurdle rates were set in line with existing regulatory standards (for example, Austra-lia, the Netherlands, and the United Kingdom as an addi-tional benchmark). In one instance, the 2019 Basel III target for common Tier 1 was applied as a supplementary bench-mark for crisis credibility purposes (that is, Spain).

The capital assessment is also influenced by the change in RWAs over the risk horizon. Economic capital models and the credit risk assumptions underpinning the credit risk treatment in the Basel II/III framework imply a positive rela-tionship between unexpected losses implied by RWAs (that is, potential worst- case losses) and default risk (and the re-sulting recovery rates). However, actual practice in stress tests for major country FSAPs has been varied:

• RWAs were kept constant (for example, China, Japan, and Mexico).

• RWA weights were kept constant, but the total RWA amounts were adjusted for credit growth and/or credit losses. This method corresponds approximately to the evolution of RWAs for banks using the standardized approach Basel II (for example, Russia, Saudi Arabia, and Spain).

• RWAs changed under stress due to changes in the risk profile, in addition to the effects from asset growth (for example, France, Germany, Japan, Luxembourg, the Netherlands, and the United Kingdom). These changes are consistent with the rules for risk weights according to the Basel framework, which are either im-plicitly captured (for example, based on information

nonrealized and/or strategic disposals (for example, loan books in runoff or sales of noncore businesses)5 and acquisitions are generally excluded. Firms are also expected to replace maturing exposures unless there is a sound basis for assuming that this will not happen (for example, deleveraging plans for banks in IMF program countries).

There is not necessarily a specific “best practice” associ-ated with each assumption on the factors that bank manage-ment controls. However, general conservatism (often aligned with historical experience) should be an important consider-ation. FSAPs have sought to ensure some uniformity in their application where possible, and to match their specific relevance to the country in question. Detailed guidance on how these assumptions should be implemented is usually provided (Appendix 15.2).

Capital Standards

The assessment of bank solvency in stress tests requires a consistent definition and appropriate measurement of capital standards, which comprise two elements: (1) the definition of capital, and (2) the calculation of capital adequacy (based on the choice of capital metric(s), hurdle rate(s), assumptions on RWAs, and the nature of data consolidation). Any capital shortfall under stress indicates the amount of potential re-capitalization. Since the definition of capital varies across jurisdictions but directly impacts the scale of capital short-fall, there needs to be full disclosure of the composition of capital (Appendix 15.3), along with sufficient information about the implications of the planned adoption of regulatory changes— for example, phasing out of some types of eligible capital (BCBS 2010c, 2010a)—over the stress test risk hori-zon. With the finalization of the second stage of Basel III in late 2017 (BCBS 2017a), the transitional arrangements con-tinue through 2027, and will have to be considered for the forthcoming stress tests (BCBS 2017b).

In FSAPs, the definition of capital is usually aligned with the one applied by country authorities. Over recent years, the capital definitions used in FSAP stress tests were guided by the jurisdictions’ implementation status of the Basel framework. All major countries have adopted Basel II (BCBS 2012b, 2012c) and, more recently, also Basel III capital stan-dards (FSB 2017).6 They either:

• Followed Basel II requirements (for example, Austra-lia, India, Indonesia, Luxembourg, Netherlands, Turkey, Russia, and the United States);

• Changed in line with the Basel III transition schedule for those that were moving to the new re-gime (for example, Brazil, France, Japan, Spain, and

5 Except when there are legally binding commitments under competition rules, for example, as agreed with the European Commission in the case of EU Member States.

6 The Basel I definition is rarely used among the major countries (except for banks in Indonesia and specific groups of banks in the United States).





378

TABLE 15.3

Original Basel III Transition ScheduleBasel II and III: Current and Phase- In Arrangements

(All dates are as of January 1)

2011 2012 2013 2014 2015 2016 2017 2018 As of Jan. 1 2019

Leverage ratio Supervisory monitoring Parallel run Jan. 1, 2013–Jan. 1, 2017

Disclosure started Jan. 1, 2015

Migration to

Pillar 1Minimum common equity capital

ratio2.0% 2.0% 3.5% 4.0% 4.5% 4.5% 4.5% 4.5% 4.5%

Capital conservation buffer 0.625% 1.250% 1.875% 2.5%Minimum common equity plus

capital conservation buffer3.5% 4.0% 4.5% 5.125% 5.750% 6.375% 7.0%

Phase- in of deductions from CET1 (including amounts exceeding the limit for DITAs, MSRs, and financials)

20.0% 40.0% 60.0% 80.0% 100.0% 100.0%

Minimum Tier 1 Capital 4.0% 4.0% 4.5% 5.5% 6.0% 6.0% 6.0% 6.0% 6.0%Minimum total capital 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0%Minimum total capital plus

conservation buffer8.0% 8.0% 8.0% 8.625% 9.25% 9.875% 10.5%

Capital instruments that no longer qualify as non- Core Tier 1 Capital or Tier 2 Capital

Phased out over 10-year horizon beginning 2013

Liquidity coverage ratio Observation period begins

Introduce minimum standard

Net stable funding ratio Observation period begins


Currently Basel II Transition to Basel IIISource: Basel Committee for Banking Supervision (BCBS).Note: See BCBS 2010a, 2010b, 2017a, and Appendix 15.3 for capital definitions. Revisions to the liquidity risk framework under Basel III (BCBS 2013) resulted in the graduated introduction of the Liquidity Coverage Ratio (LCR). Specifically, the LCR was introduced as planned on January 1, 2015, but the minimum requirement began at 60 percent, rising in equal annual increments of 10 percentage points to reach 100 percent on January 1, 2019. CET1 = Common Equity Tier 1; DTAs = deferred tax assets; MSRs = mortgage servicing rights. For the most recent transi-tional arrangements see BCBS 2017b.



to questions about the consistency and comparability of FSAP stress test results across countries and their implica-tions for the associated stability analysis.

A comprehensive FSAP solvency stress testing exercise would preferably consist of three components: a balance sheet module, a market- based portfolio model, and spillover analysis. Balance- sheet- based methods cover a wide range of risks and exposures, often based on granular, reliable data, which tend to be most accessible, and, thus, represent the core of solvency stress tests. Market- based portfolio models are better able to reflect changes in actual risks (as reflected in investor perceptions) and the dependencies across multiple risks (possibly with greater flexibility to quantify point estimates at higher levels of statistical confi-dence) (see “Stress Test Models” section). Spillover analysis, which captures contagion risk and feedback effects, has be-come an important element of solvency stress tests in in-creasingly interconnected financial systems; however, the development of robust models in this area remains nascent (for example, Espinosa- Vega and Solé 2011), in large part due to data limitations.

Macroeconomic and Satellite Modeling

System- wide stress tests that are informed by adverse macro-economic conditions affecting the profitability and solvency of banks require satellite models that help project the impact of key sources of risk. Under each scenario, these models determine how changes in macroeconomic and financial sector variables impact impairments, various income com-ponents, such as net interest income (including funding costs), noninterest income, and trading income as well as credit growth as input to the solvency stress tests (Fig-ure 15.4). Satellite models can be run at the economy level,

from the Basel Committee for Banking Supervision’s Sixth Quantitative Impact Study, such as for Germany and the United Kingdom), or are treated more explic-itly (for example, France). In other words, the evolu-tion of RWAs is determined by changes in the estimated PDs and LGDs of a firm and/or portfolio level for IRB banks, subject to the evolution of total credit exposure under stress. For some countries, im-plicit IRB risk weights were simulated, to reflect the economic risk profile of banks that are still under the standardized approach (for example, Brazil).

• RWAs for operational and market risks are often as-sumed to either (1) remain unchanged, or (2) change proportionally to the changes in RWAs for credit risk (mainly for market risk). FSAP stress tests are typically based on the assumption that the asset structure of banks remains the same during the stress test horizon, that is, there are portfolio rebal-ancing and/or substitution effects, such as replacing maturing loans with securities that attract different (usually lower) risk weights.

Going forward, projecting RWAs during times of stress would need to also reflect revised specifications in the final-ized Basel III framework as of late 2017 (BCBS 2017b).

Method

Once the key elements of the stress testing framework have been determined, various quantitative methods can be ap-plied to estimate capital adequacy under projected financial stress. However, the stress testing literature provides little guidance on the selection and application of appropriate models in different circumstances. This issue has given rise

Source: Authors.Note: P&L = profit and loss statement; RWAs = risk-weighted assets.

Figure 15.4 General Representation of Satellite Modeling in Bank Solvency Stress Testing

Stress event(applied macroeconomic shock)

Macroeconomic models(impact of stress event

on macroeconomicvariables)

Financial shocks, for example,haircuts to securities portfolios

(consistent withmacroeconomic scenarios)

Satellite models(“translation”)

Impact on bank financials(P&L, balance sheet)

Credit growth Components ofpre-provision income

Credit losses andchange in RWAs

Other, for example,dividend payout



pending on the characteristics of the respective financial sys-tem. For simple financial systems (with predominantly local banks), stress tests are normally less resource- intensive and can be completed with simpler models. In contrast, more complex systems require correspondingly more advanced stress testing methods to capture a wide range of material risks.

The stress testing methods are based on either a determin-istic or stochastic framework. Deterministic approaches are predicated on prudential information in balance- sheet- based stress test specifications, while stochastic frameworks incor-porate uncertainty around these accounting identities using historical volatility and/or market information, usually in the context of portfolio- based models. Both approaches can ac-commodate scenario and sensitivity analyses.

It is important to be aware of the differences between dif-ferent stress testing methods and their implications for the interpretation of the results. As a general rule, the more so-phisticated the model, the higher the estimation uncertainty. At the same time, simpler methods might be inadequate for highly interconnected and complex banking sectors with large credit and market risk exposures. When different ap-proaches (TD, BU) and models are used in FSAPs, the re-sults are cross- validated and the differences reconciled, which includes a discussion of the assumptions and caveats associated with the different models.

Most stress testing approaches do not fully reflect the impact of macro- financial dynamics during times of stress. For instance, existing FSAP stress tests do not adequately capture feedback effects beyond the initial impact of

sectoral level, and also at the level of individual banks (or of one of their specific portfolios).

The construction of satellite models typically comprises three key steps: (1) the choice of the estimation method, (2) the selection of the dependent variable and a set of potential ex-planatory variables that form the initial model specification, and (3) the iterative process of fitting the model (and complet-ing robustness checks).

Various types of modeling may be used. These include time series analysis, regression models (for example, ordi-nary least squares regression, logistic regression, and panel data analysis), and structural models (Foglia 2009; Drehm-ann 2009). Most major country FSAP stress tests have typi-cally relied on the country authorities’ satellite models, on the basis that these models would have undergone repeated calibrations and robustness checks over time. The FSAP team sometimes cross- validates with the IMF staff’s own sat-ellite models in parallel TD tests (see Figures 15.5 and 15.6 for application in the FSAP Update for the United Kingdom [IMF 2011d]).

Stress Test Models

Given the rapidly evolving characteristics of financial sys-tems in most countries, there is no stress testing model that is perfectly suited for a particular financial system. However, the chosen modeling approach should adequately capture the complexity, uniqueness, and idiosyncrasies of that sys-tem, subject to data availability. In FSAPs, the choice of the appropriate stress test model(s) can vary significantly de-

Source: IMF 2011d.Note: BU = bottom up; BoE = Bank of England; CPI = consumer price index; FSA = Financial Supervisory Authority; FSAP = Financial Sector Assessment Program; RAMSI = Risk Assessment Model of Systemic Institutions; TD = top down; WEO = World Economic Outlook.

Figure 15.5 Example of Satellite Model Estimations for Bank Solvency Stress Testing: 2011 UK FSAP Update

Macroeconomicscenarios

Real GDPgrowth

Short-terminterest rates

Unemployment

Inflation (CPI)

Long-terminterest rates

House priceinflation

Commercial realestate inflation

Net interestincome

Operatingexpenses

Trading income

Noninterestincome

Credit losses

Macro-financial variablesgenerated by macro

model with Bayesian updating→ inputs to both BU and

TD stress tests

FSA and BoEIMF

Idiosyncratic variablesgenerated by RAMSI model

→ inputs to both IMFsatellite models

BoE

WEO baseline

Double-dip (DD) mild

Double-dip (DD)severe

Slow growth (SG)



Macroeconomic scenarios

Satellite models CCA Components• Equity valuation and debt level• Funding cost• Market price of risk

Network model of thebanking sector

Portfolio model of bankingsector risk

System assetsloss distribution Haircuts on sovereign and bank debt

Capital ratios Joint capital shortfall

Satellite Model 1Dynamic panel

regression usingmacro andfirm data

RAMSI

Net interest income

Operating expensesTrading income

Noninterest income

Credit losses

Systemic CCA

Satellite Model 2Adjust optionformula by

net operatingprofit only

Idiosyncratic variables

Macro-financial variablesShort-term interest rates

UnemploymentInflation (CPI)

Long-term interest rates

House price inflationCRE inflation

Source: IMF 2011d.Note: CCA = contingent claims analysis; CPI = consumer price index; CRE = commercial real estate; FSAP = Financial Sector Assessment Program; RAMSI = Risk Assessment Model of Systemic Institutions.

Figure 15.6 Example of Application of Satellite Model Outputs to Top- Down Bank Solvency Stress Test Models: UK FSAP Update

macroeconomic shocks on the banking sector, notwith-standing some work in this area (for example, Vitek and Bayoumi 2011; Catalán and others 2017; Krznar and Mathe-son 2017). The literature and the actual use of stress test models that include macro- financial feedback effects remain very limited (Alfaro and Drehmann 2009). An important reason is that the interaction between adverse macroeco-nomic scenarios, such as changes in credit aggregates, and firm- level financial soundness complicates the specification of feedback effects (BCBS 2009).

The IMF staff deploys a suite of stress testing models for surveillance stress testing. They can be categorized into two broad strands, supplemented by a third, and are discussed elsewhere in this section. These approaches are not mutually exclusive in that there are overlaps in the types of data uti-lized (Table 15.4 and Figure 15.7). The IMF staff has cata-logued models developed within the institution to improve transparency in the models used in FSAPs and other areas of IMF work (Ong 2014).

Accounting- Based (balance sheet) Models This approach is most pervasive and has the longest history of use given its rela-tive simplicity (for example, simulations could be done in a spreadsheet). It has the added attraction of directly producing results in terms of regulatory variables (for example, capital ad-equacy ratios). A variety of stress testing tools for banking sector analysis at various levels of development have been deployed for FSAPs and other surveillance work and technical

assistance (Čihák 2007; Ong, Maino, and Duma 2010; Schmieder, Puhr, and Hasan 2011). The balance- sheet- based approach remains the cornerstone of FSAP stress testing and continues to be applied even in the largest, most systemic fi-nancial systems, including in all major country FSAPs to date. The network model used for spillover analysis in FSAPs ( Espinosa- Vega and Solé 2011) can also be considered an accounting- based approach.

Market- Price- Based Models The market- price- based mod-els are often built on portfolio riskmanagement techniques and typically derive concise “systemic risk measures” from estimated dependencies among different risk factors. These risks (for example, sovereign, credit, and market) are typi-cally excluded when modeling the default risk of each insti-tution in isolation (Segoviano and Padilla 2006; Gray, Jobst, and Malone 2010; Chan- Lau 2010; Gray and Jobst 2011; Jobst and Gray 2013). Unlike accounting values, risk- based measures of solvency include additional considerations to in-form the assessment of capital adequacy during times of fi-nancial stress (Figure 15.8):

• The possibility that institutions may fail simultaneously (joint default risk): Most conventional stress tests are agnostic to default dependencies across institutions, that is, when one risk factor increases the likelihood of realization of other risk factors (with common shocks affecting multiple firms at the same time), es-pecially under stressful conditions. The joint default





382

TABLE 15.4

Scorecard on Data and IMF Stress Test ModelsData Quality

Basic Data Detailed Data

Dat

a Ty

pe Balance sheet ✓ ✓ ✓ ✓ ✓ ✓

Macroeconomic ✓ ✓ ✓ ✓ ✓

Supervisory ✓ ✓ ✓ ✓

Market ✓ ✓ ✓ ✓

Interbank ✓

Stre

ss T

est E

lem

ents

Methodology Accounting- based

Accounting- based incorporating macro- financial models

Accounting- based

Market- price- based

Market- price- based

Accounting- and market- price- based incorporating macro- financial models

Accounting- based incorporating macro- financial models

Market- price- based incorporating macro- financial models

Model Balance sheet approach

Balance sheet approach

Network approach

Extreme- value theory approach

CoRisk Systemic Contingent Claims Analysis

Satellite models Distress dependence framework

Shock Sensitivity analysis

Macroscenarios Sensitivity analysis

Sensitivity analysis

Sensitivity analysis

Macroscenarios Macroscenarios Macroscenarios

Source: Ong 2014.Note: For descriptions of models, see: Espinosa- Vega and Solé 2011 for the network approach; Chan- Lau, Mitra, and Ong 2012 for the extreme value theory approach; Jobst and Gray 2013 and Gray and Jobst 2011 for Systemic Contingent Claims Analysis; Chan- Lau 2010 for CoRisk; and Segoviano and Padilla 2006 for the distress dependence framework.



served at a certain point in time reflects the outcome of a stochastic process rather than a discrete value. Given that individual (and joint) default risk varies over time (and depends on joint effects across banks as discussed above), distribution- based approaches to capital assess-ments based on the historical volatility of risk factors represent a clear conceptual departure from conven-tional balance- sheet- based stress testing techniques. Unlike RWAs, risk- based measures of solvency (such as market- implied expected losses and the correspond-ing capital shortfalls) consider the actual historical dynamics of default risk, such as value at risk or ex-pected shortfall (that is, the average density of extreme losses beyond value at risk at a selected percentile level

risk of banks within a system depends on the indi-vidual bank’s propensity to cause and/or propagate shocks due to adverse changes in one or more risk factors (a distribution- based approach). Given that large shocks are transmitted across entities differently from small shocks, measuring nonlinear dependence in stress testing can provide important insights into the joint tail risks that arise in extreme loss scenarios. This would also include measuring the differential ef-fects of the joint realization of multiple risk factors, which affects system- wide capital adequacy.

• The sensitivity of stress test results to the historical volatility of risk factors ( risk- based capital adequacy): Prudential information based purely on accounting identities ob-

Sources: Ong 2014; and authors.Note: CCA = contingent claims analysis; EVT = extreme value theory.

Figure 15.7 Stress Test Models Developed by the IMF Staff

Accounting-BasedApproach

Balance Sheet ApproachNetwork Approach

Market-Price-BasedApproach

CoRiskDefault DependenceFrameworkEVT Approach

SatelliteModels CCA

Macro-FinancialApproach

DefaultDependenceFramework

Aggr

egat

e Lo

sses

VaR95 percent

Probability

ExtremeLoss

“Tail Risk”

ES95 percentAverage densitybeyond VaR95 percent

“Distributional Approaches” (estimated or simulated

tail risk)

Value at Risk (VaR) or Expected Shortfall (ES)

VaR99%


(e.g., 1 standard

deviation)


Expected Loss = Minimum Regulatory Capital

Mean

Unexpected Loss= Economic CapitalExpected drift of aggregate losses


0

Time

T

Balance Sheet Approach(accounting values)

Source: Jobst and Gray 2013; and authors.

Figure 15.8 Key Conceptual Differences in Loss Measurements between the Accounting- Based and Market- Price- Based Approaches



reasons, which means that the design of a meaningful pre-sentation format for analysis by the FSAP team is essential (Figures 15.9–15.11). Specifically, the presentation of the ag-gregated results by the authorities should be consistent with local regulatory requirements and, where relevant, any tran-sition to a new regulatory regime (for example, Basel III). It should also be sufficiently granular to cover the presentation of the following information:

• Name of each institution or peer group (if con-strained by confidentiality) included in the stress testing exercise;

• Dispersion of capital adequacy levels (before and af-ter the application of different stress scenarios), such as the minimum, the maximum, and the interquar-tile range (for example, the 25th, 50th, and 75th per-centiles) if they are not presented by institution;

• Outcome for each year of the risk horizon;• Capital shortfall if one or more institutions fail to

meet the predefined hurdle rate of capital adequacy (in absolute terms, as a percentage of GDP, and as a percentage of total sector assets within the scope of the exercise);

• Details on the contributions of different drivers (for example, profitability, credit/trading losses, RWA) of the results; and

• Assumptions and limitations of the design and im-plementation of the stress test(s).

The findings of the stress tests are then used to (1) provide quantitative support for the FSAP’s stability risk assessment, and (2) facilitate policy discussions with the authorities on risk- mitigation strategies and crisis preparedness.

Publication

As a final step in the FSAP stress testing process, the main findings are published. The disclosure of the stress test re-sults (in addition to the standards assessment) forms a sub-stantial part of public accountability but is often a very sensitive issue (especially during times of macro- financial challenges) and requires supervisory discretion (if the effec-tive implementation of remedial actions might be compro-mised). In addition to providing a meaningful judgment on the outcome of the test (for instance, the fact that no bank fails a test does not mean that vulnerabilities do not exist), findings should be appropriately nuanced to ensure that the information does not promote a false sense of security or cause undue alarm. In FSAPs, this objective is commonly achieved through:

• Clear documentation of definitions, assumptions, models, and limitations of stress tests in Technical Notes and/or as supplementary information in the FSSA report;7

[or conditional tail expectation]). Hence, in this distribution- based approach, the capital adequacy as-sessment reflects the variability of both assets and lia-bilities at different levels of statistical confidence.

The market- based stress tests are likely to involve varying approaches of linking valuation methods to prudential stan-dards, which could make them less tractable across countries due to flexible modeling. They usually do not show direct links to key regulatory ratios, which need to be derived in separate, additional steps. Prices of certain market instru-ments (for example, equity prices and credit- default- swap spreads) as essential inputs to market- based stress tests are not always readily available. Thus, data limitations have ren-dered these approaches supplementary to accounting- based approaches in FSAPs. For instance, distress dependence models were used in only a handful of major country FSAPs where the necessary data were available for robust estimation and credible implementation (for example, Germany, Mex-ico, Spain, Sweden, the United Kingdom, and the United States), and provided interesting insights into the interlink-ages of the default risk of sample banks.

Macro- Financial Models Macro- financial models represent the third strand of stress testing approaches. They are fo-cused on the system- wide impact of different transmission channels between macroeconomic and financial conditions. By specifying certain macroeconomic situations, stress test-ers would apply consistent combinations of multiple shocks (for example, GDP, employment, inflation, exchange rate, interest rates, and asset prices), which could simultaneously affect various segments of banks’ businesses and exposures, and hence potentially result in overall capital losses. Macro- financial stress testing can be implemented by means of both accounting- and market- price- based models, by estimating additional macro- financial linkages affecting risk parame-ters used in the simulation exercises. The market- based mod-els that fall into this category include the Systemic Contingent Claims Analysis and distress dependence (due to some structural elements in the specification of changes in joint default risk), while satellite models may also be classi-fied as macro- financial in nature.

Communication

Presentation of Outputs

Stress tests are aimed at drawing attention to the significance of key vulnerabilities during times of adverse macroeco-nomic and/or market conditions and deliver findings that may trigger closer supervisory review of certain financial ac-tivities, if necessary. Thus, results need to be presented in an accessible manner to appropriately convey the findings, which would allow country authorities to draw policy rec-ommendations that inform appropriate prudential actions. In FSAPs, stress test results, especially those generated via the BU approach, are often aggregated for confidentiality

7 However, the publication of these documents is voluntary for country authorities.



likelihood of the realization of specific scenarios; (2) applying more encompassing stress tests (that is, complementary accounting- and market- price- based models) with a wider coverage of risks; and (3) ensuring a more consistent and cohe-sive presentation of assumptions and results to support an ef-fective comparability of implementation and findings across different countries.

Building on the progress so far, IMF stress tests will re-quire continuous innovation and adjustment to adequately capture relevant risks in an evolving and an exceedingly com-plex international financial system. Important areas for im-provement that will ensure that FSAP stress tests are fit for their purpose include: (1) the integration between solvency and liquidity risks; (2) spillover analysis, both within a finan-cial system and across borders; and (3) the incorporation of feedback loops between the real economy and financial sec-tor. The IMF staff has published a set of “best practice” prin-ciples on macro- financial stress testing, drawing on the accumulated experience of more than a decade of FSAPs (Chapter 2). This chapter, in turn, illustrates the application of these principles by reviewing key elements of IMF stress tests— specifically, in the major country FSAPs— during the global financial crisis and their actual implementation.

While greater harmonization of methods and approaches have helped enhance the consistency and comparability across countries, complete standardization of FSAP stress tests remains elusive. Qualitative analysis and expert judg-ment are, and will continue to be, indispensable for what amounts to an art form rather than an exact science. Given the many “moving parts” of stress tests, sufficient flexibility not only ensures that their design and implementation re-main relevant amid a constantly evolving spectrum of risks but also helps absorb innovative methodologies to adequately capture them— subject to considerable variations in local regulatory requirements and the political sensitivities affect-ing the communication of stress test results. That said, these challenges should not undermine the value of well- designed stress tests.

• Mandatory summaries of the stress testing exercises in a standardized format (that is, the STeM) as part of the FSSA to improve transparency and facilitate cross- country comparisons (Table 15.5); and

• Disclosure of the aggregated results of stress tests af-ter conclusion of the FSAP, including a minimum amount of information such as the relevant poststress ratio(s) and the respective amount(s) of capital shortfall.8

In the case of the FSAPs covered in this chapter, all sample countries have published their FSSAs. In almost all cases, Technical Notes on the respective stress testing exercises were completed (with the exception of Australia and Spain, where the details are described in appendices to the respective FSSAs), but only a few countries (that is, Germany, Sweden, the United Kingdom, and the United States) consented to their publica-tion (see IMF 2010c, 2011b, and 2011a, respectively).

CONCLUSIONThe lead- up to the global financial crisis illustrated that sur-veillance stress tests are not fail- safe, stand- alone diagnostic tools. Conceptually, the implementation of stress tests is very challenging due to: (1) diverse business models and ac-tivities of sample banks; (2) varying degrees of estimation uncertainty associated with models, based on assumptions that may not be sufficiently robust to capture all the relevant risks; (3) binding constraints to data availability and quality; and (4) prudential concerns and/or political sensitivities af-fecting the formulation of credible stress scenarios. The com-plexity of running stress tests is magnified during crises when rapidly changing financial conditions and heightened market expectations require a carefully planned communi-cation strategy.

At the IMF, significant efforts continue to be made to ad-dress the identified shortcomings. Some of the steps taken in-clude: (1) standardizing the shock scenarios across countries, where possible, and making nascent attempts to quantify the

8 Country authorities rarely agree to the publication of the results of indi-vidual banks.





386

Figure 15.9 Example of Bottom- Up Bank Solvency Stress Test Output Template Provided to Banks: UK FSAP Update1

Macroscenario [select] Baseline Double Dip (mild) Double Dip (severe) Slow Growth

23456789101112

15161718192021222324

Inpu

t sys

tem

-wid

e ag

greg

atio

n

[Bank Name]

Main Results

Hurdle RateAssumption

Prestress(End-2010)

Y1(2011)

Y2(2012)

Y3(2013)

Y4(2014)

Y5(2015)

Failed stress test? (1 = yes, 0 = no)Total Capital

Tier 1Common Equity Tier 1

Capital needs to recapitalize bankIn GBP millions

Total CapitalTier 1


Capital needs to recapitalize bank (relative to total assets)In percent

Total CapitalTier 1


1

Hurdle Rate Total Capital 8.0% 8.0% 8.0% 8.0% 8.0%4.0% 4.0% 4.5% 5.5% 6.0%2.0% 2.0% 3.5% 4.0% 4.5%

Hurdle Rate Tier 1 CapitalHurdle Rate Common Equity Tier 1

25

Not p

ublic

ly re

porte

d Sensitivity Check I:Like "Main Results" but without Basel III

capital phase-in/phase-out and

capital grandfathering





Total CapitalTier 1



Total CapitalTier 1


14



2728293031323334353637

40414243444546474849

Not p

ublic

ly re

porte

d Sensitivity Check II:Like "Main Results"

but with capital buffer(see higher hurdle

rates below)





Total CapitalTier 1



Total CapitalTier 1


26



5051

Risk Drivers

Net profit (before losses)Credit lossesOverall trading/valuation losses

Losses from sovereign debt holdings - trading book & AfSLosses from financial sector debt holdings - trading book & AfSLosses from sovereign debt holdings - held-to-maturityLosses from financial sector debt holdings - held-to-maturityLosses from FX shock

RWAsNet profit (before losses)Credit lossesOverall trading/valuation losses

Losses from sovereign debt holdings - trading book & AfSRisk Drivers

39


Andreas A

. Jobst, Li Lian Ong, and C

hristian Schmieder

387

2.0%

2.0%

2.0%

2.0%

Inpu

t sys

tem

-wid

e ag

greg

atio

n

thereof (if applicable): counterparty credit riskthereof (if applicable): securitization in banking book

Change of market risk RWAsthereof (if applicable): stressed value at riskthereof (if applicable): equity standard measurement methodthereof (if applicable): incremental risk charge and securitization in trading book

Change of operational risk RWAs

Stress testparameters(In percent)

5960616263646566In

put s

yste

m-w

ide

aggr

egat

ion

Background

Total capital adequacy ratio (In percent)

Common equity Tier 1 ratio (In percent)Tier 1 capital ratio (In percent)

Total capital

Common equity Tier 1 capitalTier 1 capital

Leverage (capital/assets)

Dividend yield (dividend paid/equity) (In percent)Return on total regulatory capital (In percent)

58

69707172737475767778798081828384

68 Percentage of profits retainedPhase-in of deductions from Core Tier 1 capitalPhase-out of non-Tier 1 and Tier 2 capitalCredit risk

PD/NPL ratio (average)LGD (average)Asset correlation (average)Credit growth

Asset risk-weightingsChange of credit risk RWAs

353637

40414243444546474849




50515253545556

Inpu

t sys

tem

-wid

e ag

greg

atio

n Risk Drivers





Change in RWAs (In percent)

Risk Drivers(In percent of RWAs)

39

Source: Authors.Note: Results should be reported for hurdle rate assumptions without capital buffers (lines 10–12). Alternative stress test results should be based on hurdle rates that either ignore the capital phase-in/phase-out provisions of Basel III (lines 14–22) or include a capital buffer (lines 26–34). These results, however, have no impact on other sections of the main stress test and serve merely as a basis for sensitivity analysis. All other results reported in the spreadsheet (from line 39 onward) are based on the main results obtained from a stress testing set-up consistent with the Basel III treatment of capital but without capital buffers (lines 1–9). AfS = available for sale; FSAP = Financial Sector Assessment Program; FX = foreign exchange; GBP = UK Pound Sterling; LGD = loss given default; PD/NPL = probability of default/nonperforming loan; RWAs = risk-weighted assets. 1The actual template is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit.

Figure 15.9 (continued)






388


23456789101112

15161718192021222324

Repo

rted

as s

ampl

e av

erag

e fo

r all

firm

s

[Bank Name]

Main Results


Prestress(end-2010)

Y1(2011)

Y2(2012)

Y3(2013)

Y4(2014)

Y5(2015)




Total CapitalTier 1



Total CapitalTier 1


1



25

Not p

ublic

ly re

porte

d Sensitivity Check I:Like "Main Results" but without Basel III

capital phase-in/phase-out and

capital grandfathering





Total CapitalTier 1



Total CapitalTier 1


14



2728293031323334353637

40414243444546474849

Not p

ublic

ly re

porte

d Sensitivity Check II:Like "Main Results"

but with capital buffer(see higher hurdle

rates below)





Total CapitalTier 1



Total CapitalTier 1


26



5051

Risk Drivers(Sum of all firms)




Losses from sovereign debt holdings - trading book & AfSRisk Drivers

39 Figure 15.10 Example of Bottom- Up Bank Solvency Stress Test Output Template Provided to Authorities: UK FSAP Update1


Andreas A


hristian Schmieder

389

353637

40414243444546474849




50515253545556

Repo

rted

as s

ampl

e av

erag

e fo

r all

firm

s

Risk Drivers(Sum of all firms)





Change in RWAs (In percent)

Risk Drivers(In percent of RWAs)

39

2.0%

2.0%

2.0%

2.0%

Repo

rted

as s

ampl

e av

erag

e fo

r all

firm

s

thereof (if applicable): counterparty credit riskthereof (if applicable): securitization in banking book

Change of market risk RWAsthereof (if applicable): stressed value at riskthereof (if applicable): equity standard measurement methodthereof (if applicable): incremental risk charge and securitization in trading book

Change of operational risk RWAs

Stress testparameters(In percent)

5960616263646566Re

porte

d as

sam

ple

aver

age

for a

ll firm

s

Background

Total capital adequacy ratio (In percent)

Common equity Tier 1 ratio (In percent)Tier 1 capital ratio (In percent)

Total capital

Common equity Tier 1 capitalTier 1 capital

Leverage (capital/assets)

Dividend yield (dividend paid/equity) (In percent)Return on total regulatory capital (In percent)

58

69707172737475767778798081828384

68 Percentage of profits retainedPhase-in of deductions from Core Tier 1 capitalPhase-out of non-Tier 1 and Tier 2 capitalCredit risk

PD/NPL ratio (average)LGD (average)Asset correlation (average)Credit growth

Asset risk-weightingsChange of credit risk RWAs

Source: Authors.Note: Results should be reported for hurdle rate assumptions without capital buffers (lines 10–12). Alternative stress test results are based on either hurdle rates that ignore the capital phase-in/phase-out provisions of Basel III (lines 14–22) or include a capital buffer (lines 26–34). These results, however, have no impact on other sections of the main stress test and serve merely as a basis for sensitivity analysis. All other results reported in the spreadsheet (from line 39 onward) are based on the main results obtained from a stress testing set-up consistent with the Basel III treatment of capital but without capital buffers (lines 1–9). AfS = available for sale; FSAP = Financial Sector Assessment Program; FX = foreign exchange; GBP = UK Pound Sterling; LGD = loss given default; PD/NPL = probability of default/nonperforming loans; RWAs = risk-weighted assets.1The actual template is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit.

Figure 15.10 (continued)






390

Source: Authors.Note: FSAP = Financial Sector Assessment Program; ppts = percentage points.1The actual template is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit.

Figure 15.11 Example of Bottom- Up Bank Solvency Stress Test Summary Template Provided to Authorities: UK FSAP Update1


[Bank Name] Prestress(end-2010)

Y1(2011)

Y2(2012)

Y3(2013)

Y4(2014)

Y5(2015)

10Distribution of Tier 1 Capital

25507590

Count

> –3 ppts

3 ppts

By ratio

Percentile

10Distribution of Common Equity Tier 1 Capital

25507590

Count

> –3 ppts

3 ppts

By ratio

Percentile

–2 ppts–1 pptRegulatory Minimum1 ppt2 ppts

Percentile Percentile Percentile Percentile



10Distribution of Total Capital

25507590

Count

> –3 ppts

3 ppts

By ratio

Percentile






TABLE 15.5

Example of Stress Test Matrix (STeM) for Bank Solvency Risk: Spain FSAP UpdateDomain Assumptions

Top Down by Authorities Top Down by FSAP TeamInstitutions included • Commercial banks and intervened

savings banks• All publicly listed banks with sufficient pricing

historyMarket share • Over 96 percent of the banking

sector, excluding foreign branches• About 45 percent of the banking sector, excluding

foreign branchesData and baseline date • Supervisory data as at end- 2011

• Scope of consolidation: legal entity as at end- 2011

• Risk horizon of two years, under crisis conditions

• Publicly available market and statutory data. Scope of consolidation: legal entity as at end- 2011

• Risk horizon of two years, under crisis conditions

Methodology (for example, included in scenario analysis linking solvency and liquidity, separate test using ad hoc model/balance sheet)

• BdE macro- financial panel regression model (estimates capital shortfall) without behavioral adjustments

• IMF balance sheet approach (estimates capital shortfall)

• Systemic CCA model (estimates expected losses, capital shortfall, and contingent liabilities)

Risks (for example, funding liquidity shock, market liquidity shock, both)

• “ Double- dip” recession (severe and short- term) scenario of one standard deviation from the IMF- projected baseline GDP growth trend over a two- year risk horizon— without positive adjustment dynamics toward the end of the (short) risk horizon

• The second, more adverse scenario further escalates the macroeconomic shock by increasing the shock to two- year real GDP growth by another 2.5 percentage points

• Sovereign risk reflected in valuation haircut to AfS and trading book debt holdings• Extra provisioning and capital add- on due to regulatory changes

Regulatory standards • Basel II transitioning to Basel III and Basel III capital requirements slightly exceeded (4 percent Core Tier 1 hurdle rate for both years)

• Basel III capital definition• RWAs remain constant but are subject to changes due to deleveraging by banks in both 2012

and 2013Results • Postshock, more than a third of all banks in the system would not be able to comply with Basel

III hurdle requirements until end- 2013 irrespective of the choice of top- down model• The BdE model reveals projected impairment losses of around € 73 billion under the IMF

adverse scenario, which generates capital shortfall of about € 18 billion compared with a Core Tier 1 capital hurdle rate of 4 percent

• Based on the Systemic CCA results, challenges exist from the realization of low probability tail risk of multiple firms experiencing a dramatic escalation of losses. In the IMF adverse scenario, the largest (and publicly listed) banks would experience a market- implied capital shortfall of more than € 14 billion on average (with a peak in excess of € 21 billion at end- 2012) at a statistical probability of 5 percent or less (expressed as “tail risk”)

Source: IMF 2012b.Note: AfS = available for sale; BdE = Banco de España; Systemic CCA = Systemic Contingent Claims Analysis; FSAP = Financial Sector Assess-ment Program; RWAs = risk- weighted assets.


Appendix 15.1.FSAP Solvency

Stress Tests FY2010–FY13





394


Financial Sector Assessment Program Solvency Stress Tests FY2010–FY13: Stress Test Matrix for S- 25 and Other G20 Countries1

United States2

Indonesia3

China4

Luxembourg5

Netherlands6

Germany7

United Kingdom8

Turkey9

RussiaTiming of FSAP FY 2010 FY2010 FY2011 FY2011 FY2011 FY2011 FY2011 FY2011 FY2011

Stress Testing Framework 1. Scope

Approach Bottom-up • No. • Bottom-up (BU) by banks, in

collaboration with authorities and IMF.

• BU by banks in collaboration with authorities and IMF.

• No. • No. • No. • BU by banks in collaboration with authorities and IMF.

• BU by banks in collaboration with authorities and IMF.

• BU by banks, in collaboration with authorities and IMF.

Top-down • 3 top-down (TD) tests by IMF.

• TD by IMF in collaboration with authorities.

• TD by authorities.• TD by IMF.

• TD by authorities. • TD by IMF in collaboration with authorities.


• TD by IMF.

• TD by authorities.• TD by IMF.


• TD by authorities.

Coverage

Institutions • 54 bank holding companies (BHCs) using balance sheet (B/S) approach.

• 36 BHCs using the Consistent Information Multivariate Density (CIMDO) methodology.

• 14 Systemically Important Financial Institutions using Systemic Contingent Claims Analysis (SCCA).

• TD: All 121 commercial banks, excl. rural banks (115 for scenario analysis; all for sensitivity analysis).

• BU: 12 largest banks (8 for scenario analysis, all for sensitivity analysis).

• 17 banks (5 large commercial, 12 joint-stock commercial).

• 108 subsidiaries and branches. • 7 banks. • 3 banking groups, 16 largest German banks (14 SIFIs plus two Landesbanken), the savings banks (Sparkassen), and the other cooperative banks; very small banks were excluded from the sample.

• 14 SIFIs (SCCA).

• 6 largest banks + largest building society.

• 9 largest banks. • BU: 15 largest banks.• TD: All commercial banks (1,012).

Market share • B/S: 85 percent.• CIMDO: 59 percent.• SCCA: 70 percent.

• TD: 100 percent.• BU: 60 percent.

• 66 percent of total banking sector assets (86 percent of commercial banking sector assets).

• 100 percent. • 85 percent. • B/S: 85 percent.• SCCA: 60 percent.

• 88 percent. • Over 80 percent. • BU: 56 percent.• TD: 100 percent.

Reporting basis • Consolidated banking groups. • Unconsolidated banking groups. • Consolidated banking groups. • Unconsolidated local entities. • Consolidated banking groups. • Unconsolidated domestic businesses. • Consolidated banking groups. • Unconsolidated domestic businesses.

• Unconsolidated local entities.

Data

Source • Publicly available data. • BU: Banks’ own data.• TD: Supervisory and publicly available

data.

• BU: Banks’ own data.• TD by authorities: Supervisory and

publicly available data.• TD by IMF: Publicly available data.

• Supervisory data. • Supervisory data. • Supervisory and publicly available data.

• BU: Banks’ own data.• TD: Supervisory and publicly

available data.

• BU: Banks’ own data.• TD: Supervisory data.

• BU: Banks’ own data.• TD: Supervisory data.

Cut-off date • End-2009. • As at Sep 2009. • End-2009. • As at Jun 2010. • As at Jun 2010. • Hybrid:— End-2009 for B/S positions.— Sep 2010 for P&L.

• End-2010. • End-2010. • End-2010.

2. Scenario Design

Risk horizon • 2010-14 (5 years). • 2009Q4-2012Q4 (3 years).

• Scenario: 2010 (1 year).• Sensitivity: 1Q, 1 year or 2 years.

• 2011-12 (2 years). • 2011-15 (5 years). • 2011-15 (5 years). • 2011-15 (5 years).

• BU by banks: 2011 (1 year).• TD: 2011-13 (3 years): Sudden stop.• TD: 2011-14 (4 years): Boom and bust.

• Instantaneous. • 2011 (1 year).

Scenarios Baseline • WEO Apr 2010. • WEO Apr 2009. • N/A. • WEO Oct 2010. • WEO Oct 2010. • WEO Oct 2010. • WEO Oct 2010. • WEO Feb 2011. • Slightly below WEO Jan 2011.

Growth shocks (calculated per Committee of European Banking Supervisors for Scenario Designs unless indicated otherwise)

• Combined impact of four adverse shocks: (i) sizeable and persistent shock to growth rate of potential output; (ii) an additional short run demand shock, reflecting high unemployment, weak credit, and continued fall in housing prices; (iii) further near-term fiscal stimulus to support near-term growth; and (iv) rising inflation expectations.

• Output gap falls by 2.3 percentage points relative to baseline in adverse scenario.

• Output gap falls by 3.3 percentage points relative to baseline in alternative adverse scenario.

• ≈ 1/3 output loss experienced during Asian crisis.

• GDP growth down from 12 percent to: —7 percent (mild) —5 percent (medium) —4 percent (severe).

• 1 standard deviation (SD). • 1 SD.• 2 SD.

• 1.5 SD (1 SD and 2 SD run by FSAP team for internal comparisons).

• 1 SD.• 2 SD.

• Sharp contraction over four quarters followed by a sluggish recovery over the next 12 quarters.

• A two-year boom in growth and credit followed by a sharp contraction over four quarters and then a sluggish recovery.

• 1 SD.• 1.7 SD.

Slow growth scenario • Yes. • No. • Yes • No. • No. • Yes. • Yes. • Yes. • No.

Risks

Key risk(s) • Credit risk. • Credit risk. • Credit risk associated with rapid loan growth.

• Credit risk. • Credit risk. • Credit risk. • Credit risk. • Credit risk. • Credit risk.• Adjustments for regulatory

forbearance.

Other risks covered in scenario analysis

• N/A. • N/A. • N/A. • Sovereign risk, in both trading and banking books (CEBS model).

• Sovereign risk, in both trading and banking books (CEBS model).

• Off-balance sheet exposures.

• Sovereign risk in trading book only (IMF models); application of sovereign haircuts on banking book in sensitivity analysis completed separately by IMF staff.

• Sovereign and banking risks in both trading and banking books (IMF models).

• Funding risk.

• N/A. • Sovereign and other debt holdings, in trading book and AfS in banking book.

• Propagation channel through network effects.

• Liquidity stress measured by its solvency impact (losses from fire sales of liquid assets)

Other tests/risks • Sensitivity tests: Credit and market risks.

• Sensitivity tests: Credit and market risks.

• Sensitivity tests: Credit and market risks, including: (i) largest individual exposures; (ii) real estate sector exposures; (iii) exposures to local government financing platforms (LGFPs); (iv) exposures to overcapacity industries; and (v) exposures to export sectors.

• Contagion risk.• Reverse stress test.



• N/A • N/A • Sensitivity tests: Credit and market risks.

• Spillover risk through domestic network effects (included in macro scenarios).

Source: Compiled by authors with contributions from respective FSAP stress testers.Note: The IMF fiscal year runs from May 1 to April 30. The table presented here is a representation only; a full- sized version is available as an MS Excel® file on the IMF eLibrary at www.elibrary.imf.org/page/stress -test2-toolkit. AfS = available for sale; CEBS = Committee of European Banking Supervisors; SIFI = systemically important financial institution.


www.elibrary.imf.org/page/stress-test2-toolkit. AfS = available for sale; CEBS = Committee of European Banking Supervisors; SIFI = systemically important financial institution

www.elibrary.imf.org/page/stress-test2-toolkit. AfS = available for sale; CEBS = Committee of European Banking Supervisors; SIFI = systemically important financial institution

Appendix 15.2.Example of Summary of Key

Assumptions Applied in Solvency Stress Testing





396


Example of Summary of Key Assumptions Applied in Solvency Stress Testing Exercise: UK FSAP UpdateDomain Element Specific Rules/Assumptions(Risk) factors

assessedLoss ratesProfitabilityFixed income holdingsFX shockTaxes

• Credit losses based on satellite models developed by firms depending on scenario. • Profit (interest income, interest expenses, net fee, and commission income, and operating expenses) should be based on firm’s satellite

models (or expert judgment). For end- 2010, net profit before tax should be adjusted for extraordinary income/losses to avoid misleading results.

• Trading income based on satellite model or statistical matching of both trading income and GDP growth using a parametric fit of their historical distribution (for example, a decline in GDP growth is assumed to reduce trading income).

• Funding costs based on satellite model for interest expenses, including a nonlinear effect. Changes in funding costs due to different solvency conditions cannot be smaller than the one generated by either some general funding cost sensitivity or results from suggested CCA- based approach (Appendix 15.3, Option 2). These changes are unaffected by possible balance sheet deleveraging.

• Mark- to- market impact on fixed- income holdings: Focuses on the projection of haircuts for holdings of both sovereign and bank debt based on IMF approach. These haircuts will be applied to both trading and banking book.

• Sovereign and financial sector debt holdings: Haircut on holdings in the banking and trading books based on market expectations over five years after controlling for changes of market valuation during 2010 as developed by the IMF staff. Cash at central banks, repos, and asset swaps where there is no economic interest in the security (for instance, instruments held against assets pledged to the Bank of England) are excluded. Moreover, haircuts are applied only to issuers that are non-“AAA” rated.

• FX shock: Firms are asked to report separately the marginal impact of the following FX shock of the following currencies on net open positions: US dollar, euro, and Japanese yen. The shock for each currency should be twice the standard deviation of the respective FX volatility during 2010 and impact the trading book in 2011 (100 percent) and 2012 (50 percent) only.

• Tax assumption: 25 percent in case of positive profits, zero otherwise.Behavioral

adjustment of banks

Dividend payout rules (similar to Basel III minima)

Credit growthAsset disposalCapital raising

• Balance sheets are assumed to be constant (that is, they grow in line with nominal GDP).• Dividend payout depends on capitalization under stress: dividend payout only if firm reports profits over the past year; if total capital

ratio is above 8 percent (after the envisaged dividend payout and, at the same time, exhibits sufficient Tier 1 and common equity Tier 1 capitalization) but below the 10.5 percent threshold (which reflects the magnitude of the proposed “capital conservation buffer” under Basel III), the firm is considered capital- constrained and needs to follow a defined payout schedule.

• Credit growth in line with nominal GDP for banks with a Tier 1 capital buffer of 2.5 percentage points above the regulatory minimum (that is, hurdle rate); credit growth decreases by 2 percentage points for each decrease in Tier 1 capital by 1 percentage point once the capital buffer is less than 2.5 percentage points above the Tier 1 capital hurdle rate. Hence, growth becomes negative when capitaliza-tion is at the minimum capital ratio unless nominal GDP grows by more than 5 percent.

• Other business strategy considerations: Asset disposals or acquisitions over time should not be considered, except where legally binding commitments under EU State aid rules exist. Maturing exposures are assumed to be replaced. Any interim capital raising until end- 2010 can be considered in calculations.

Source: IMF 2011d.Note: CCA = contingent claims analysis; FX = foreign exchange rate; FSAP = Financial Sector Assessment Program.


Appendix 15.3.Example of Comparison Table on

Relevant Core Tier 1 Capital Definitions





398


Example of Comparison Table on Relevant Core Tier 1 Capital Definitions: UK FSAP UpdateCapital Component Basel II Basel III EBA FSA General Prudential Sourcebook

FSA Interim Capital Regime/FSAP Bottom- Up Stress Testing

Core Tier 1 (CT1)

• Ordinary shares.• Retained earnings and

reserves.• Share premium

account.• Minority interests.

• Ordinary shares.• Retained earnings and reserves.• Share premium account

relating to CT1 instruments.• Minority interests

(subject to limits).

• Ordinary shares.• Retained earnings and

reserves.• Share premium account

relating to CT1 instruments.

• Minority interests.

• Ordinary shares. • Retained earnings and reserves.• Share premium account

relating to CT1 instruments.• Minority interests.

• Ordinary shares. • Retained earnings and reserves.• Share premium account

relating to CT1 instruments.• Minority interests.

• Externally verified interim net profits.

• Interim net profits. • Externally verified interim net profits.

• Existing government support measures counted as CT1.



Core Tier 1 Filters

• Existing national filters - see FSA GENPRU column for UK filters.

• Pension deficit net of DRA (if approach chosen).

• Unrealized gains on AfS equities.

• Unrealized gains on Investment property.

• Unrealized gains on land and buildings.

• Unrealized losses (gains) on AfS debt.

• Pension deficit net of DRA (if approach chosen).

• Unrealized gains on AfS equities.

• Unrealized gains on investment property.

• Unrealized gains on land and buildings.

• Unrealized losses (gains) on AfS debt.

• Cash- flow hedge reserve not fair valued on balance sheet.

• Gain on sale related to securitization transactions.

• Cumulative gains and losses due to changes in own credit risk on fair valued financial liabilities.








Andreas A


hristian Schmieder

399

APPENDIX TABLE 15.3.1 (continued)Example of Comparison Table on Relevant Core Tier 1 Capital Definitions: UK FSAP UpdateDeductions

from Core Tier 1

• Interim net losses. • Interim net losses.• Intangibles including goodwill

(limited recognition of mortgage servicing rights).

• Interim net losses.• Intangibles including

goodwill.

• Interim net losses. • Interim net losses.• Intangibles including

goodwill.

• Investments in own shares.• Shortfall of the stock of

provisions to expected losses.

• Investments in own shares.• 50 percent shortfall in stock

of provisions to expected losses.

• 50 percent of certain securitization exposures.

• Investments in own shares.• 50 percent shortfall in stock

of provisions to expected losses.

• 50 percent of certain securitization exposures.

• Defined benefit pension fund assets and liabilities (include liabilities in full, deduct assets).

• Deferred tax assets (limited recognition allowed).

• Reciprocal cross holdings in the common stock of banking, financial, and insurance entities.

• Investments in the common stock of banking, financial, and insurance entities that are outside the scope of regulatory consolidation and where the bank does not own more than 10 percent of the issued common share capital.

• Significant investments in the common stock of banking, financial, and insurance entities that are outside the scope of regulatory consolidation (limited recognition).

• Certain qualifying holdings.

• 50 percent material holdings in financial institutions (excluding material insurance holdings).

• 50 percent free deliveries.

Source: IMF 2011d.Note: AfS = available for sale; DRA = deficit reduction amount; EBA = European Banking Authority; FSA = Financial Supervisory Authority; FSAP = Financial Sector Assessment Program; GENPRU = general pruden-tial sourcebook for banks, building societies, insurers, and investment firms.



Requirement.” BCBS Publication 445, Bank for International Settlements, Basel, Switzerland, December. https://www.bis.org /bcbs/publ/d445.htm.

Bernanke, Ben S., 2010. “The Supervisory Capital Assessment Program— One Year Later.” Speech delivered at the 46th Annual Conference on Bank Structure and Competition, Federal Re-serve Bank of Chicago, Chicago, May. http://www.federalreserve .gov/newsevents/speech/bernanke20100506a.htm.

Board of Governors of the Federal Reserve System. 2009. “The Su-pervisory Capital Assessment Program: Overview of Results.” Board of Governors of the Federal Reserve System, Washing-ton, DC, May.

———. 2012a. “Comprehensive Capital Analysis and Review.” Board of Governors of the Federal Reserve System, Washington, DC, March. http://www.federalreserve.gov/bankinforeg/ccar.htm.

———. 2012b. “Comprehensive Capital Analysis and Review 2012: Methodology for Stress Scenario Projection.” Board of Governors of the Federal Reserve System, Washington, DC, March. http://www.federalreserve.gov/newsevents/press/bcreg /20120312a.htm.

———. 2012c. “Supervisory Guidance on Stress Testing for Banking Organizations with More Than $10 Billion in Total Consolidated Assets.” Board of Governors of the Federal Re-serve System, SR Letter 12-7, Washington, DC, May.

Board of Governors of the Federal Reserve System, Federal De-posit Insurance Corporation, and Office of the Comptroller of the Currency. 2012. “Guidance on Stress Testing for Banking Organizations with Total Consolidated Assets of More Than $10 Billion.” SR Letter 12–7, Board of Governors of the Federal Reserve System, Washington, DC, May. https://www .federalreserve.gov/supervisionreg/srletters/sr1207.htm.

Borio, Claudio, Mathias Drehmann, and Kostas Tsatsaronis. 2012, “ Stress- testing Macro Stress Testing: Does It Live Up to Expectations?” BIS Working Paper 369, Bank for International Settlements, Basel, Switzerland, January. http://www.bis.org /publ/work369.htm.

Catalán, Mario, Dale Gray, Laura Valderrama, and TengTeng Xu. 2017. “ Macro- Financial Feedbacks in Stress Testing.” Presenta-tion at Joint IMF- EBA Colloquium— New Frontiers on Stress Testing, London, March 1.

Central Bank of Ireland. 2011. “The Financial Measures Programme Report.” Central Bank of Ireland, Dublin, Ireland, March. https://www.centralbank.ie/publication/financial-measures -programme.

Cerutti, Eugenio, and Christian Schmieder. 2012. “The Need for “ Un- consolidating Consolidated Banks’ Stress Tests.” IMF Work-ing Paper 12/288, International Monetary Fund, Washing-ton, DC. https://www.imf.org/en/Publications/WP/Issues/2016 /12/31/ The- Need- for- Un- consolidating- Consolidated- Banks - Stress- Tests-40151.

Chan- Lau, Jorge. 2010. “Regulatory Capital Charges for Too- Connected- To- Fail Institutions: A Practical Proposal.” IMF Working Paper 10/98, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/ Regulatory- Capital- Charges- for- Too- Connected - to- Fail- Institutions- A- Practical- Proposal- 23753.

———, Srobona Mitra, and Li Lian Ong. 2012. “Identifying Contagion Risk in the International Banking System: An Ex-treme Value Theory Approach.” International Journal of Finance and Economics 17 (4): 390–406.

Čihák, Martin. 2007. “Introduction to Applied Stress Testing.” IMF Working Paper 07/59, International Monetary Fund,

REFERENCESAlfaro, Rodrigo, and Mathias Drehmann. 2009. “Macro Stress Tests

and Crises: What Can We Learn.” BIS Quarterly Review (Decem-ber): 29–41. https://www.bis.org/publ/qtrpdf/r_qt0912e.htm.

Banco de España. 2012. “Bank Recapitalisation and Restructuring Process: Results of the Independent Evaluation of the Spanish Sector: Presentation by the Deputy Governor.” http://www.bde .es/bde/en/secciones/prensa/infointeres/reestructuracion /valoracionesind/.

Basel Committee on Banking Supervision (BCBS). 2009. “Princi-ples for Sound Stress Testing Practices and Supervision.” BCBS Publication 155, Bank for International Settlements, Basel, Switzerland, May. https://www.bis.org/publ/bcbs155.htm.

———. 2010a. “Basel III: A Global Regulatory Framework for More Resilient Banks and Banking Systems.” BCBS Publica-tion 189, Bank for International Settlements, Basel, Switzer-land, December. http://www.bis.org/publ/bcbs189.htm.

———. 2010b. “Basel III: International Framework for Liquidity Risk Measurement, Standards and Monitoring.” BCBS Publi-cation 188, Bank for International Settlements, Basel, Switzer-land, December. https://www.bis.org/publ/bcbs188.htm.

———. 2010c. “Results of the Comprehensive Quantitative Impact Study.” BCBS Publication 186, Bank for International Settlements, Basel, Switzerland, December. http://www.bis .org/publ/bcbs186.htm.

———. 2011. “Global Systemically Important Banks: Assessment Methodology and the Additional Loss Absorbency Require-ment.” BCBS Publication 207, Bank for International Settle-ments, Basel, Switzerland, November. http://www.bis.org /publ/bcbs207.htm.

———. 2012a. “Peer Review of Supervisory Authorities’ Implementation of Stress Testing Principles.” BCBS Publication 218, Bank for International Settlements, Basel, Switzerland, April. http://www.bis.org/publ/bcbs218.htm.

———. 2012b. “Report to G20 Leaders on Basel III Implementa-tion.” BCBS Publication 220, Bank for International Settle-ments, Basel, Switzerland, June. http://www.bis.org/publ/bcbs220 .htm.

———. 2012c.“Progress Report on Basel III Implementation.” BCBS Publication 232, Bank for International Settlements, Basel, Switzerland, October. https://www.bis.org/publ/bcbs232 .htm.

———. 2012d. “A Framework for Dealing with Domestic Sys-temically Important Banks.” BCBS Publication 233, Bank for International Settlements, Basel, Switzerland, October. https://www.bis.org/publ/bcbs233.htm.

———. 2013. “Basel III: The Liquidity Coverage Ratio and Li-quidity Risk Monitoring Tools.” BCBS Publication 238, Bank for International Settlements, Basel, Switzerland, January. http://www.bis.org/publ/bcbs238.htm.

———. 2017a. “Basel III: Finalising Post- crisis Reforms.” BCBS Publication 424, Bank for International Settlements, Basel, Switzerland, January. https://www.bis.org/bcbs/publ/d424.htm.

———. 2017b. “Basel III Transitional Arrangements, 2017-2027.” Bank for International Settlements, Basel, Switzerland, December.

———. 2018a. “Stress Testing Principles.” BCBS Publication 450, Bank for International Settlements, Basel, Switzerland, Octo-ber. https://www.bis.org/bcbs/publ/d428.htm.

———. 2018b. “Global Systemically Important Banks: Revised Assessment Methodology and the Higher Loss Absorbency




http://www.federalreserve.gov/newsevents/speech/bernanke20100506a.htm

http://www.federalreserve.gov/newsevents/speech/bernanke20100506a.htm

http://www.federalreserve.gov/bankinforeg/ccar.htm.



https://www.federalreserve.gov/supervisionreg/srletters/sr1207.htm

https://www.federalreserve.gov/supervisionreg/srletters/sr1207.htm






http://www.bde.es/bde/en/secciones/prensa/infointeres/reestructuracion/valoracionesind/




















https://www.imf.org/en/Publications/WP/Issues/2016/12/31/The-Need-for-Un-consolidating-Consolidated-Banks-Stress-Tests-40151



https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Regulatory-Capital-Charges-for-Too-Connected-to-Fail-Institutions-A-Practical-Proposal-23753




Howard, Stacia. 2008. “Stress Testing with Incomplete Data: A Practical Guide.” In Proceedings of the IFC Conference on Mea-suring Financial Innovation and Its Impact. Basel: Bank for International Settlements. www.bis.org/ifc/publ/ifcb31.htm.

International Monetary Fund (IMF). 2010a. “Integrating Stability Assessments under the Financial Sector Assessment Program into Article IV Surveillance.” IMF Policy Paper, Washington, DC, August. https://www.imf.org/en/Publications/Policy-Papers /Issues/2016/12/31/ Integrating- Stability- Assessments- Under - the- Financial- Sector- Assessment- Program- into- PP4477.

———. 2010b. “Integrating Stability Assessments under the Finan-cial Sector Assessment Program into Article IV Surveillance— Background Material.” IMF Policy Paper, Washington, DC, August. https://www.imf.org/en/Publications/Policy-Papers/Issues /2016/12/31/ Integrating- Stability- Assessments- Under- the - Financial- Sector- Assessment- Program- into- PP4478.

———. 2010c. “United States: Technical Note on Stress Testing.” IMF Country Report 10/244, Washington, DC, July. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ United - States- Publication- of- Financial- Sector- Assessment- Program - Documentation- Technical- 24101.

———. 2010d. “The IMF- FSB Early Warning Exercise: Design and Methodological Toolkit.” Washington, DC, August https://www.imf.org/en/Publications/Policy-Papers/Issues /2016/12/31/The-IMF-FSB-Early-Warning-Exercise-Design -and-Methodological-Toolkit-PP4479.

———. 2011a. “Germany: Technical Note on Stress Testing.” IMF Country Report 11/371, Washington, DC, December. https://www.imf.org/en/Publications/CR/Issues/2016/12 /31/ Germany- Technical-Note-on-Stress-Testing-25461.

———. 2011b. “Sweden: Financial Sector Assessment Program Update— Technical Note on Contingent Claims Analysis Ap-proach to Measure Risk and Stress Test the Swedish Banking Sec-tor.” IMF Country Report 11/286, Washington, DC, September. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / Sweden- Financial- Sector- Assessment- Program- Update- Technical - Note- on- Contingent- Claims- 25241.

———. 2011c. “United Kingdom: Financial System Stability As-sessment.” IMF Country Report 11/222, Washington, DC, July. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / United- Kingdom- Financial- System- Stability- Assessment-25111.

———. 2011d. “United Kingdom FSAP Update: Stress Testing the Banking Sector Technical Note.” IMF Country Report 11/227, Washington, DC, July. https://www.imf.org/en/Publications /CR/Issues/2016/12/31/ United- Kingdom- Stress- Testing- the - Banking- Sector- Technical- Note- 25119.

———. 2012a. “ Macro- financial Stress Testing: Principles and Practices.” IMF Policy Paper, Washington, DC, August.

———. 2012b. “Spain: Financial System Stability Assessment.” IMF Country Report. 12/137, Washington, DC, June. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Spain - Financial- System- Stability- Assessment- 25977.

———. 2013. “Mandatory Financial Stability Assessments under the Financial Sector Assessment Program: Update.” IMF Policy Paper, Washington, DC, November. http://www.imf.org/en /publications/ policy- papers/issues/2016/12/31/mandatory - financial- stability- assessments- under- the- financial- sector - assessmentprogram- pp4838.

———. 2018. “ IMF- FSB Early Warning Exercise.” IMF Fact Sheet, Washington, DC. March 8. https://www.imf.org/en /About/Factsheets/Sheets/2016/08/01/16/29/ IMF- FSB- Early - Warning- Exercise.

Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/ Introduction- to- Applied- Stress- Testing- 20222.

Committee of European Banking Supervisors (CEBS). 2010. “2010 EU Wide Stress Testing.” July. London: Committee of European Banking Supervisors.

Drehmann, Mathias. 2009. “Macroeconomic Stress- testing Banks: A Survey of Methodologies.” In Stress Testing the Banking Sys-tem: Methodologies and Applications, edited by Mario Quagliari-ello. New York: Cambridge University Press.

Espinosa- Vega, Marco A., and Juan Solé. 2011. “ Cross- border Financial Surveillance: A Network Perspective.” Journal of Fi-nancial Economic Policy 3 (33): 182–205.

European Banking Authority (EBA). 2010. “Aggregate Outcome of the 2010 EU-wide Stress Test Exercise Coordinated by CEBS in Cooperation with the ECB.” European Banking Au-thority, London, July. https://www.eba.europa.eu/documents /10180/15938/Summaryreport.pdf/95030af2-7b52-4530-afe1 -f067a895d163.

———. 2011a. “2011 EU-Wide Stress Testing Exercise.” Euro-pean Banking Authority, London, March.

———. 2011b. “Overview of the 2011 EU-Wide Stress Test: Meth-odological Note.” European Banking Authority, London, October. https://eba.europa.eu/documents/10180/26923/Sovereign-capital -shortfall_Methodology-FINAL.pdf/acac6c68-398e-4aa2-b8a1 -c3dd7aa720d4.

Fell, John. 2006. “Overview of Stress Testing Methodologies: From Micro to Macro.” Presentation to the Korea Financial Supervisory Commission/Financial Supervisory Service- International Mone-tary Fund Seminar on Macroprudential Supervision Conference on Challenges for Financial Supervisors, Seoul, November.

Financial Services Authority (FSA). 2009. “Stress and Scenario Testing” Policy Statement PS09/20, London, December.

———. 2011. “Prudential Risk Outlook.” Financial Services Au-thority, London, March. http://www.fsa.gov.uk/pages/library /corporate/pro/index.shtml.

Financial Stability Board (FSB). 2011. “Policy Measures to Ad-dress Systemically Important Financial Institutions.” Financial Stability Board, Basel, Switzerland, November. http://www .fsb.org/2011/11/r_111104bb/.

———. 2012. “Extending the G- SIFI Framework to Domestic Systemically Important Banks.” Progress Report to the G- 20 Finance Ministers and Central Bank Governors, Financial Sta-bility Board, Basel, Switzerland, April. http://www.fsb.org /2012/04/r_120420b/.

———. 2017. “Implementation and Effects of the G20 Financial Regulatory Reforms: Third Annual Report.” Progress Report to G20 Leaders, Financial Stability Board, Basel, Switzerland, July.

Foglia, Antonella. 2009. “Stress Testing Credit Risk: A Survey of Authorities’ Approaches.” International Journal of Central Bank-ing 5 (3): 9–45. http://www.ijcb.org/journal/ijcb09q3a1.htm.

Gray, Dale F., and Andreas A. Jobst. 2011. “Modelling Systemic Financial Sector and Sovereign Risk.” Sveriges Riksbank Eco-nomic Review (2): 68–106.

Gray, Dale F., Andreas A. Jobst, and Samuel Malone. 2010. “Quantifying Systemic Risk and Reconceptualizing the Role of Finance for Economic Growth.” Journal of Investment Manage-ment 8 (2): 90–110.

Hardy, Daniel C., and Christian Schmieder, 2013. “Rules of Thumb for Bank Solvency Stress Testing.” IMF Working Paper 13/232, International Monetary Fund, Washington, DC. https://www .imf.org/en/Publications/WP/Issues/2016/12/31/ Rules - of- Thumb- for- Bank- Solvency- Stress- Testing- 41047.


http://www.bis.org/ifc/publ/ifcb31.htm

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/






https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/The-IMF-FSB-Early-Warning-Exercise-Design-and-Methodological-Toolkit-PP4479.











https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Spain-Financial-



http://www.imf.org/en/publications/


https://www.imf.org/en/About/Factsheets/Sheets/2016/08/01/16/29/




https://www.eba.europa.eu/documents/10180/15938/Summaryreport.pdf/95030af2-7b52-4530-afe1-f067a895d163.



https://eba.europa.eu/documents/10180/26923/Sovereign-capital-shortfall_Methodology-FINAL.pdf/acac6c68-398e-4aa2-b8a1-c3dd7aa720d4



http://www.fsa.gov.uk/pages/library/corporate/pro/index.shtml

http://www.fsa.gov.uk/pages/library/corporate/pro/index.shtml

http://www.fsb.org/2011/11/r_111104bb/.

http://www.fsb.org/2011/11/r_111104bb/.

http://www.fsb.org/2012/04/r_120420b/.

http://www.fsb.org/2012/04/r_120420b/.













Moretti, Marina, Stéphanie Stolz, and Mark Swinburne. 2008. “Stress Testing at the IMF.” IMF Working Paper 08/206, International Monetary Fund, Washington, DC. https://www .imf.org/en/Publications/WP/Issues/2016/12/31/ Stress- Testing - at- the- IMF- 22275.

Ong, Li Lian, editor. 2014. A Guide to IMF Stress Testing: Methods and Models. Washington, DC: International Monetary Fund.

Ong, Li Lian, Rodolfo Maino, and Nombulelo Duma. 2010. “Into the Great Unknown: Stress Testing with Weak Data.” IMF Working Paper 10/282, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/ Into- the-Great-Unknown-Stress-Testing-with - Weak- Data- 24488.

Schmieder, Christian, Claus Puhr, and Maher Hasan. 2011. “Next Generation Balance Sheet Stress Testing.” IMF Working Paper 11/83, International Monetary Fund, Washington, DC. https://w w w.imf.org/en/Publ icat ions/W P/Issues/2016/12/31 / Next- Generation- Balance- Sheet- Stress- Testing- 24798.

Segoviano, Miguel, and Pablo Padilla. 2006. “Portfolio Credit Risk and Macroeconomic Shocks: Applications to Stress Testing under Data- Restricted Environments.” IMF Working Paper 06/283, In-ternational Monetary Fund, Washington, DC. https://www.imf .org/en/Publications/WP/Issues/2016/12/31/ Portfolio- Credit - Risk- and- Macroeconomic- Shocks- Applications- to- Stress - Testing- Under- Data- 20064.

Vitek, Francis, and Tamim Bayoumi. 2011. “Spillovers from the Euro Area Sovereign Debt Crisis: A Macroeconometric Model Based Analysis.” CEPR Discussion Paper 8497, Centre for Eco-nomic Policy Research, London. https://ideas.repec.org/p/cpr /ceprdp/8497.html.

International Monetary Fund, Financial Stability Board, and Bank for International Settlements. 2009. “Guidance to Assess the Systemic Importance of Financial Institutions, Markets and In-struments: Initial Considerations.” Report to the G- 20 Finance Ministers and Central Bank Governors, October.

International Monetary Fund (IMF) and World Bank. 2003. “Ana-lytical Tools of the FSAP.” International Monetary Fund, Wash-ington, DC, February. http://www.imf.org/external/np/fsap/2003 /022403a.htm.

Jobst, Andreas A., and Dale F. Gray. 2013. “Systemic Contingent Claims Analysis— Estimating Market- Implied Systemic Solvency Risk.” IMF Working Paper13/54, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/Systemic-Contingent- Claims- Analysis - Estimating- Market- Implied- Systemic- Risk-40356.

Jobst, Andreas A., and Hiroko Oura. Forthcoming. “Sovereign Risk— Haircut Estimation and Bank Solvency Stress Testing.” IMF Working Paper, International Monetary Fund, Washington, DC.

Jobst, Andreas A., Li Lian Ong, and Christian Schmieder. 2013. “A Framework for Macroprudential Bank Solvency Stress Test-ing: Application to S- 25 and Other G20 Country FSAPs.” IMF Working Paper 13/68, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/ A- Framework- for- Macroprudential- Bank- Solvency - Stress- Testing- Application- to- S- 25- and- Other- G- 40390.

Krznar, Ivo, and Troy Matheson. 2017. “Towards Macroprudential Stress Testing: Incorporating Macro Feedback Effects.” IMF Working Paper 17/149, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2017/06/30/Towards-Macroprudentia l-Stress-Test ing - Incorporating- Macro- Feedback- Effects- 44955.











https://ideas.repec.org/p/cpr/ceprdp/8497.html

https://ideas.repec.org/p/cpr/ceprdp/8497.html

http://www.imf.org/external/np/fsap/2003/022403a.htm

http://www.imf.org/external/np/fsap/2003/022403a.htm

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Systemic-Contingent-Claims-




https://www.imf.org/en/Publications/WP/Issues/2017/06/30/Towards-Macroprudential-Stress-Testing-Incorporating-









CHAPTER 16

Macroprudential Liquidity Stress Testing in FSAPs for Systemically Important Financial Systems

ANDREAS A. JOBST • LI LIAN ONG • CHRISTIAN SCHMIEDER

Bank liquidity stress testing, which has become de rigueur following the costly lessons of the global financial crisis, remains underdeveloped com-pared to solvency stress testing. The ability to adequately identify, model, and assess the impact of liquidity shocks, which are infrequent but

can have a severe impact on financial systems, is complicated not only by data limitations but also by interactions among multiple factors. This chapter provides a conceptual overview of liquidity stress testing approaches for banks and discusses their implementation by the IMF staff in the Financial Sector Assessment Program for countries with systemically important financial sectors between 2010 and 2016.

funding maturity and currency mismatches at the time. The postcrisis regulatory reforms widened the prudential perime-ter to encourage better liquidity risk management.

This chapter provides a conceptual overview of liquidity stress testing approaches developed by the IMF staff and sur-veys the staff’s application to assessing system- wide vulnerabili-ties to market and funding liquidity risks. In particular, it focuses on the Financial Sector Assessment Programs (FSAPs) in countries with systemically important financial sectors. It is the companion work to Jobst, Ong, and Schmieder 2013, which reviews the IMF’s system- wide bank solvency stress test-ing. In keeping with the mandate of the FSAP, the chapter fo-cuses on the bilateral surveillance of bank liquidity risk for macroprudential purposes, that is, the extent to which disrup-tions to banks’ liquidity management result in system- wide vulnerabilities. The information in this chapter complements an internal guidance note on liquidity stress testing for the IMF staff (Catalán 2015) and includes cross- country comparisons.1

Consistent with market and regulatory developments, liquid-ity stress testing has become a core element of financial stability

1. INTRODUCTIONThe global financial crisis underscored the critical impor-tance of sound liquidity risk management for individual fi-nancial institutions, and consequently, for overall financial stability. A defining characteristic of the crisis was the simul-taneous and widespread dislocation in funding markets, which uncovered the weaknesses in banks’ liquidity profiles, particularly their increased reliance on short- term wholesale funding and high levels of leverage. Funding weaknesses were rapidly propagated through a highly interconnected global financial system, triggering contagion across financial institutions and systems and amplifying solvency concerns (IMF 2008, 2010a, 2011a).

In the wake of the crisis, the focus of the financial industry and country authorities rapidly turned to the shortcomings in liquidity riskmanagement practices. The now- obvious vul-nerabilities had been, for the most part, undetected leading up to the crisis. In hindsight, the omission could be attributed to a general lack of understanding (compared to the more famil-iar solvency risk) of— and hence insufficient attention to—

This chapter is based on IMF Working Paper 17/102 (Jobst, Ong, and Schmieder 2017). The authors would like to thank Chikako Baba, Mario Catalán, Martin Čihák, Hee- Kyong Chon, Silvia Iorgova, Michael Lau, Sylwia Nowak, Steven Phillips, Jay Surti, and Yunhui Zhao as well as the national authorities of Canada, P.R. China, Finland, India, Japan, Korea, Poland, Sweden, the United Kingdom, and the United States for their helpful comments and sugges-tions. The views expressed in this chapter do not represent those of the authors’ current employers.1 The guidance note is centered on the implementation of a stress testing methodology that focuses on the time structure of contractual cash flows. While the

guidance note is an internal document for use by the IMF staff, the current version is also available upon request.


Macroprudential Liquidity Stress Testing in FSAPs for Systemically Important Financial Systems404

last Review of the FSAP (IMF 2014a) explicitly examines systemic effects encompassing the interac-tion of different risk types within and across the vari-ous financial sectors.2

The chapter considers FSAPs undertaken in 34 significant jurisdictions that completed an FSAP exercise between Septem-ber 2010 and December 2016. This group comprises: (1) 29 countries identified by the IMF as having systemically impor-

analysis in FSAPs, which has historically focused on solvency stress testing. In doing so, the IMF staff has taken steps to:

• take stock of liquidity riskmanagement practices (for example, IMF 2010a, 2011a);

• improve liquidity stress tests by examining gaps in their previous design (for example, Ong and Čihák 2010; Schmieder and others 2012; Schmitz 2015; Jobst 2017);

• develop methods to identify systemic liquidity risk (for example, IMF 2011a; Jobst 2014); and

• build models linking liquidity and solvency risks for more robust stress tests (for example, Basel Commit-tee for Banking Supervision 2013b, 2015); the IMF’s

TABLE 16.1

S-29 and Other G20 Countries: FSAPs over the FY 2011 to FY 2017 PeriodRank Jurisdiction Grouping Completed FSAPs

since FY2010Reference

1 United Kingdom S-25/S-29*, G20, G7 2011, 2016 IMF (2011d), IMF (2016b)2 Germany S-25/S-29*, G20, G7 2011, 2016 IMF (2011h), IMF (2016c)3 United States S-25/S-29*, G20, G7 2010, 2015 IMF (2010c), IMF (2015c)4 France S-25/S-29*, G20, G7 2012 IMF (2013f)5 Japan S-25/S-29*, G20, G7 2012 IMF (2012c)6 Italy S-25/S-29*, G20, G7 2013 IMF (2013h)7 Netherlands S-25/S-29* 2011 IMF (2011b)8 Spain S-25/S-29* 2012 IMF (2012b)9 Canada S-25/S-29*, G20, G7 2014 IMF (2014c)

10 Switzerland S-25/S-29* 2014 IMF (2014e)11 China S-25/S-29*, G20 2010 IMF (2011f)12 Belgium S-25/S-29* 2013 IMF (2013d)13 Australia S-25/S-29*, G20 2012 IMF (2012g)14 India S-25/S-29*, G20 2013 IMF (2013c)15 Ireland S-25/S-29* 2016 IMF (2016e)16 Hong Kong SAR S-25/S-29* 2014 IMF (2014d)17 Brazil S-25/S-29*, G20 2012 IMF (2013e)18 Russian Federation S-25/S-29*, G20 2011, 2016 IMF (2011g), IMF (2016d)4

19 Korea S-25/S-29*, G20 2014 IMF (2015a)20 Austria S-25/S-29* 2013 IMF (2014b)21 Luxembourg S-25/S-29* 2011 IMF (2011c)22 Sweden S-25/S-29* 2011 IMF (2011e)23 Singapore S-25/S-29* 2014 IMF (2013i)24 Turkey S-25/S-29*, G20 2012 IMF (2012d), IMF (2017b)25 Mexico S-25/S-29*, G20 2012 IMF (2012a)3

26 Denmark S-25/S-29* 2014 IMF (2014f)27 Finland S-25/S-29* 2010, 2016 IMF (2010e), IMF (2017a)28 Norway S-25/S-29* 2015 IMF (2015d)29 Poland S-25/S-29* 2013 IMF (2013g)

Argentina G20 20131 IMF (2016a)European Union G20 20132 IMF (2013b)Indonesia G20 2010 IMF (2010d)Saudi Arabia G20 2011 IMF (2012f)South Africa G20 2014 IMF (2015b)

Sources: IMF 2010b; IMF 2013a; and authors. Note: See http://www.imf.org/external/np/fsap/fssa.aspx (IMF 2017c) for published FSAP country reports. S-29 coun-tries are ranked according to the size and interconnectedness of their financial systems. The IMF’s fiscal year (FY) runs from May 1 of the previous year to April 30 of the current year. FSAP = Financial Sector Assessment Program; G7 = Group of Seven; G20 = Group of Twenty; S-25 = Systemic-25 jurisdictions; S-29 = Systemic-29 jurisdictions.*Four additional countries (Denmark, Finland, Norway, Poland) were added to the original S-25 list following the 2013 decision of the IMF’s Executive Board (IMF 2014a). 1Publication delayed until February 2016. 2Stress tests were not conducted for the 2012/13 European Union FSAP. 3No separate liquidity stress test.4Liquidity stress test integrated in solvency stress test in the 2011 FSAP.

2 See also IMF 2015e and 2018.3 Note that Hong Kong SAR is not an independent country but part of

China; however, it was included in the sample due to its classification as a jurisdiction with a systemically relevant financial sector.


http://www.imf.org/external/np/fsap/fssa.aspx


4 Detailed information on the scope and specifications of the FSAP exer-cises may serve as a useful reference for these purposes, together with the existing guidance note for staff (Catalán 2015).

tant financial systems ( S- 29), which are subject to mandatory assessments every five years (IMF 2010b, 2013a, 2014a);3 and (2) the five G20 members that are not among the S- 29, as presented in Table 16.1. However, the sample for this chapter excludes three of these 34 FSAPs for which liquidity stress tests were not undertaken (European Union, Luxembourg, Mexico).

In reviewing the general concepts underpinning liquidity stress tests and their implementation in FSAPs, the chapter:

• provides the rationale (and introduces the conceptual underpinnings) for liquidity stress testing, including the regulatory framework established in recent years as well as general challenges of liquidity stress testing;

• defines a framework for system- wide liquidity stress testing by introducing a taxonomy of the main build-ing blocks— the scope, data requirements, methodol-ogy, and final output— to classify and compare the various approaches applied to FSAPs;

• reviews the parameters adopted in past FSAPs for the systemically important financial systems based on a comprehensive, cross- country Stress Test Matrix (STeM), which reflects the extent to which countries have elected to disclose the methodology and find-ings of the stress testing exercise in the respective Fi-nancial System Stability Assessment (FSSA) reports and accompanying Technical Notes on Stress Testing (Appendix 16.1, Appendix Table 16.1.1); and

• provides publicly available information on liquidity stress testing to help country authorities prepare for future FSAPs and readers seeking to develop their own stress testing framework.4

While some standardization of FSAP liquidity stress tests can improve comparability and efficiency, it may not be possible or desirable to do so under all circumstances. In fact, FSAP liquidity stress tests are far more heterogeneous than solvency stress tests. There are several reasons for this:

• Each financial system has its own special features, which also require qualitative assessment and consequently, ex-pert judgment that can influence the stress test.

• The availability and quality of data influence the choice of appropriate methods in ensuring the reli-ability and credibility of the results.

• The extent of the collaboration with the authorities (and individual banks) plays a crucial role.

The chapter is organized as follows. Section 2 sets out the premise for running liquidity stress tests and discusses the conceptual underpinnings. Section 3 details the various components and elements of the liquidity stress testing framework and their application to individual FSAPs. The caveats to liquidity stress tests are presented in Section 4. Section 5 concludes with a discussion on advances in liquid-ity stress tests and areas for future improvement.

2. WHY STRESS TEST FOR LIQUIDITY RISK?Premise

The global financial crisis provided a stark reminder of how the realization of liquidity risks can undermine financial sta-bility and underscored the need for regular liquidity stress tests on banks and banking sectors. Unlike bank solvency, which tends to deteriorate gradually during times of stress, liquidity shocks can manifest rapidly as demonstrated in the scale and scope of their impact across financial systems dur-ing the crisis. This has made liquidity risk a central element of postcrisis regulatory reforms; notably, the last revisions to the Basel Accord (“Basel III”) contain a strong emphasis on liquidity riskmanagement practices (Basel Committee for Banking Supervision 2017c). These practices comprise both quantitative (a range of metrics) and qualitative (related to risk management and supervision) measures, as documented in Table 16.2.

Liquidity stress tests inform a comprehensive assessment of whether banks’ own internal resources (in the form of li-quidity buffers) are sufficient to withstand adverse shocks. They aim to shed light on the potential need for emergency liquidity assistance to viable banks. Parent banks represent another important external source of liquidity support dur-ing times of stress, although any assessment of their capacity to do so may be limited if they are in another jurisdiction (or supervisory guidance on ring- fencing restricts cross- border transfers).

Concept

Liquidity stress tests aim to capture the risk that a bank fails to generate sufficient funding to satisfy short- term payment obligations arising from a sudden realization of liabilities. The tests assess the adequacy of available funding sources over a defined stress horizon. These tests would usually— and appropriately— examine the resilience of individual banks or banking sectors without considering central banks’ lender- of- last- resort liquidity support (Box 16.1).5 There are two broad, mutually reinforcing types of liquidity risk (Fig-ure 16.1 and Appendix 16.2):

• Funding liquidity risk is the risk that a bank will not be able to meet its current and future cash- flow needs in case of a runoff of its funding liabilities, contingent payment obligations, and/or disruptions to cash in-flows. Specifically, a bank’s funding capacity depends on whether it can manage scheduled and unsched-uled cash outflows (including the loss of funding sources and contingent lending through existing commitments) against cash inflows that are related to

5 However, most liquidity stress tests assume that in a systemic crisis, part of the eligible collateral would be pledged to the central bank (subject to the applicable haircut) for secured funding via repos as part of the open market operations (Chailloux and Jobst 2010).



Box 16.1. Central Banks and Parent Banks as Liquidity Backstops

Central banks can counterbalance rising funding risk during times of stress by acting as a lender of last resort to banks. For instance, during the global financial crisis, the US Federal Reserve entered into swap agreements with several central banks, which, in turn, provided much needed US dollar funding to their own domestic banks.1 These facilities were extended twice, enabling the European Central Bank, for instance, to provide unlimited three- month US dollar funding. In addition, the European Central Bank’s own longer- term refinancing operation program provided stable funding to eligible banks and removed intermittent funding problems during the European sovereign debt crisis.

There is also contingent liquidity support within banking groups. Parent banks could maintain or increase credit lines to subsidiaries or branches during stress periods. However, ring- fencing could hinder cross- border liquidity flows, as occurred during the global financial crisis (Cerutti and others 2010), but in the case of branches and subsidiaries in Central and Eastern Europe, funding by their Western parent banks turned out to be more reliable than alternative funding sources (for example, euro wholesale markets). Historically, parent banks typically have not provided additional liquidity to subsidiaries when they are affected by idiosyncratic liquidity shocks as a result of severe (perceived) solvency problems.

1 The US Federal Reserve also provided liquidity to large international banks (in addition to the domestic US financial institutions), but only to the US branches of foreign banking organizations.

maturing assets, the rollover risk stemming from any maturity mismatches, and the ability to access unse-cured retail/wholesale funding markets.

• Market liquidity risk is the risk that a bank will not be able to sell a sizeable volume of securities without impacting the prevailing market price (IMF 2015f). Market liquidity is reflected in volume (for example, turnover ratios) and price- based measures ( bid- ask spreads, price impact of large trades). For liquidity (and partly also solvency) stress tests, banks assess the expected cash inflows from asset sales in a stressed environment. This involves mark- to- market valuation changes (of securities that are classified as

either trading or available for sale) and possible ex-traordinary impairment losses of held- to- maturity assets in the banking book from a defaulting obligor or the forced (discounted) sale of assets by the bank (prior to the maturity date). The decline in asset val-ues owing to market risk, and the extent to which assets are subject to haircuts when used as collateral for wholesale funding, influence the severity in changes of cash flows.

The self- reinforcing downward liquidity spiral during the global financial crisis underscored the potentially crippling relationship between both types of liquidity risk (Fig-ure 16.1). The repricing of risk occurs when market illiquidity

TABLE 16.2

Liquidity Risk: Regulatory Initiatives on Liquidity RiskInitiatives Related DocumentsA. Basel Committee on Banking Supervision (BCBS) • Established the Working Group on Liquidity to review liquidity supervision of national authorities and transposed

some basic principles of liquidity risk management into standard liquidity ratios, the liquidity coverage ratio (LCR) and the net stable funding ratio (NSFR).

BCBS (2008a)

• Issued guidance on liquidity risk-management processes around 17 principles, focusing on medium and large complex banks.

BCBS (2008b)

• Developed Principles for Sound Stress Testing Practices and Supervision based on review of supervisory authorities’ implementation of stress testing principles, which integrated liquidity risk in the formulation of stress testing frameworks of banks.

BCBS (2012b)

• Proposed minimum liquidity standards via two quantitative measures (LCR and NSFR), complemented by other monitoring tools to be applied at a global level under the Basel III rules.

BCBS (2010a, 2012a, 2013a, and 2014)

• Issued guidance on the design of proposed monitoring indicators for intraday liquidity management with the aim of enabling bank supervisors to monitor banks’ intraday liquidity risk management and their ability to meet payment and settlement obligations in a timely manner and even under stressed scenarios.

BCBS (2013b)

• Issued consultative document on Stress Testing Principles, aimed at replacing existing principles published in 2009 with governing principles.

BCBS (2017d)

• Reviewed regulatory consistency of the national rules with the Basel framework (including liquidity standards) in its jurisdictional assessments (published in annual monitoring reports).

BCBS (2018)

B. Bank for International Settlements Research Task Force • Surveyed existing industry and supervisory practices in liquidity stress testing with a view to improving method-

ologies and practices, in particular, with respect to the interaction with solvency and contagion stress testing.BCBS (2013c)

• Surveyed existing literature on risk drivers of liquidity stress consistent with categories and concepts of LCR. BCBS (2013d) • Outlined several approaches to modeling the interaction between liquidity and solvency risks from a

macroprudential perspective.BCBS (2015)

• Surveyed the impact assessment of liquidity requirements and their interaction with capital requirements. BCBS (2016b)

Sources: BCBS; and authors.



UnsecuredFunding

SecuredFunding

Central BankTenders, standing facilities/discount windows

Money Markets and DepositsInterbank lending, commercial paper,

term and demand deposits

Central Bankn/a

Money MarketsRepo, securities borrowing and lending,

asset-backed commercial paper

Vulnerability Funding Source

Operational

ContingencyLiquidity Risk

Increases with higher leverage andmaturity mismatch

Capital MarketsAsset-backed securitization, covered bonds

Capital MarketsPlain vanilla debt, (private) equity, hybrid capital

Structural

Sources: Jobst 2012; and authors.

Figure 16.1 Conceptualization of Liquidity Risk

turns into funding illiquidity, such as when banks refuse to accept withdrawals. Funding illiquidity can also lead to market illiquidity, such as when swap markets dried up in late 2007 due to concerns over the rising counterparty credit risk of European banks seeking US dollar funding (IMF 2008). Both funding and market liquidity risks characterize liquidity stress tests and differentiate them from solvency stress tests. The latter assess the capital impact of asset price shocks from valuation losses and impairments that are not directly triggered by adverse funding conditions, although there is a close link to liquidity though this channel.

The proper identification, monitoring, and mitigation of both market and funding liquidity risks require both price- and quantity- based information on bank balance sheets, monetary dynamics, and developments in securities and funding markets (Table 16.3). Institution- level funding vul-nerabilities are captured by liquidity ratios, including the share of noncore funding ( short- term, wholesale, foreign ex-change) in total liabilities. These indicators are normally supplemented with a detailed decomposition of assets and liabilities, for example the share of high- quality liquid assets (HQLAs) in total assets (see Appendix 16.3), asset- liability maturity mismatches, and gross open currency positions. The assessment of liquidity risk based on this information varies with the development of monetary and general market conditions, for example, interbank market turnover, securi-ties issuance, or the volume of secured/unsecured borrow-

ing. For small open economies, trends in short- term capital inflows through financial institutions (as captured in posi-tions and flows of other investments and portfolio invest-ments received by banks) are often important indicators of noncore funding and can represent sources of instability in the funding market (Nier and others 2014).

The potential buildup of systemic vulnerabilities warrants comprehensive monitoring of liquidity risks, especially where the impact of disruptions to funding markets could be most widespread (Jobst 2014). These risks are related to different funding sources that determine the time dimension of li-quidity risk, such as the balance between secured/un-secured funding sources via capital markets and the more structural ( bank- specific) aspects of asset- liability manage-ment. Money markets and deposit funding represent short- term funding channels that meet operational requirements, while central bank money via standing facilities and tenders help reduce funding contingencies (mostly overnight and over very short time periods). Meanwhile, traditional depos-its still form the funding backbone of many banks, so liquidity risk relating to deposits also needs to be part of the risk framework.

Liquidity stress tests follow either a cumulative or a noncu-mulative approach in identifying liquidity shortfalls. The eco-nomic importance of inflows and outflows for the liquidity position of a bank or banking sector under stress can be as-sessed either in terms of a cumulative effect during a specified



syncratic in nature and difficult to capture in any model.

• Limited availability of requisite granular data (for example, asset encumbrance levels, information on collateral), and the existence of available or existing repurchase agreements (“repos”) (or reverse repos) and/or the confidentiality of bank liquidity informa-tion have constrained the development of compre-hensive liquidity stress testing models.

Funding liquidity risk has been a specific focus of recent system- wide stress tests. For instance, the 2011 and 2014 EU- wide solvency stress tests conducted by the European Banking Authority included a shock to the cost of funding, which was linked to the bank- specific impact of sovereign stress. It was also assumed that higher short- and long- term interest rates as well as lower collateral values would increase banks’ wholesale and retail funding needs (without changes to their funding structure under stress) in both the baseline and adverse scenarios. In the latter case, an explicit funding volume shock was simulated as part of the European Central Bank’s macroeconomic stress testing framework.

3. A FRAMEWORK FOR BANK LIQUIDITY STRESS TESTINGLiquidity stress testing has become a core element of the IMF’s financial sector surveillance since the global financial crisis. It is now a regular component of the financial stability module of FSAP exercises undertaken by the IMF staff (Table 16.4). Also, many countries have adopted compre-hensive approaches to assessing system- wide liquidity condi-

survival period (using implied- cash- flow [ICF] tests) or non-cumulatively by means of a limit system (such as liquidity ra-tios and associated minimum requirements). Both approaches share the common objective of capturing the risk that a bank or banking sector fails to generate sufficient funding to satisfy short- term payment obligations. Key benchmarks are two standard liquidity metrics introduced under Basel III— the liquidity coverage ratio (LCR) and the net stable funding ratio (NSFR) (see Appendix 16.3 for further information).

More comprehensive macroprudential stress tests should, where possible, incorporate negative feedback loops between solvency conditions and liquidity risk to support a more nu-anced assessment of potential systemic risk and differentiate across banks’ varying susceptibility of solvency- induced li-quidity stress. While solvency stress tests examine the impact of credit and market risk- related losses on bank capital, they would ideally also account for diminishing funding opportu-nities and the price impact of rising counterparty risk under stress, particularly in the wake of a significant deterioration of solvency conditions (see Appendix 16.4 for examples of re-search on this issue). Empirical evidence suggests that sol-vency and liquidity stress tests that do not account for the interaction between solvency and liquidity shocks substan-tially underestimate the risk exposure of individual banks and banking sectors (Puhr and Schmitz 2014). However, the practical implementation of this concept in liquidity stress testing remains at an early stage (BCBS 2013c, 2015).

The design and calibration of scenarios for liquidity stress tests tend to be more challenging than for the solvency ones because of the following factors:

• Liquidity crises are partly attributable to psychologi-cal factors or confidence effects, which tend to be idio-

TABLE 16.3

Overview of Liquidity Indicators Quantities Prices

Monetary conditions and capital flows

Base money and broader monetary aggregates Policy and money market interest ratesAccess to central bank liquidity(for example, bidding volumes)

Monetary conditions index1

Excess bank reservesVolume of short-term capital inflows(especially if intermediated by banks)

Institutional and funding liquidity

Volume of secured/unsecured funding via securities financing transactions (SFTs)2

Spread between secured/unsecured wholesale funding rate and effective policy rate

Liquidity ratios (LCR, NSFR, loan-to-deposit ratio, share of noncore funding, liquid asset ratio)

Unsecured lending rate and counterparty risk (for example, LIBOR and LIBOR-OIS spread)

Maturity mismatch measures Valuation haircuts on collateral for SFTs2

Net cash flow estimates FX swap basisGross open foreign currency position Violation of arbitrage conditions

(for example, bond-CDS basis and covered interest parity)Spreads between assets with similar credit characteristics Qualitative surveys of funding conditions

Market liquidity Volume of securities issuance Bid-ask spreads on selected assetsTransaction volumes(including average transaction size)

Qualitative fund manager surveys

Sources: Committee on the Global Financial System (CGFS) 2011; Jobst 2012; and Nier and others 2014. Note: CDS = credit default swap; FX = foreign exchange; LCR = liquidity coverage ratio; NSFR = net stable funding ratio.1Such as the elasticity of aggregate demand to the real short-term interest rate and the real effective exchange rate.2Includes repo and securities lending. See also IMF 2015f, CGFS 2011, and Markets Committee 2016 for recent studies on market liquidity.



TABLE 16.4

A Framework for Macroprudential Bank Liquidity Stress TestingComponent Description1. Scope Approach • Bottom-up (BU) by banks (using supervisory templates/assumptions; guidance from authorities/IMF staff)

• Top-down (TD) by authorities (own assumptions/templates, possibly aligned with assumptions in IMF TD stress test)• TD by authorities (IMF templates/assumptions)• TD by IMF staff (IMF templates/assumptions)

Coverage Institutions Mostly the largest banks, including foreign subsidiaries and branches Market share In most countries >80 percent of total banking sector assets Data Source Banks’ own data, supervisory data, and public data Cutoff date End-quarter or end of last fiscal year Reporting basis Mostly consolidated banking groups, but also unconsolidated domestic businesses/solo basis in many countries2. Scenario Design Test(s) • Implied-cash-flow test (cumulative/noncumulative) over 5/30 days with focus on the sudden, sizeable withdrawal of

funding (liabilities), and the sufficiency of existing assets to withstand those shocks under stressed conditions after taking into account valuation haircuts to liquid assets and amortization of outstanding assets; alternative scenarios: (1) restricted run-off to deposit and wholesale funding (that is, selected customer deposits are unaffected), (2) availability of intergroup funding, and (3) unexpected cash outflows and drawdown of unused credit lines (behavioral cash flows) due to withdrawal of contingent liabilities and inability to roll over maturing unsecured wholesale funding

• Asset-liability mismatch analysis over different risk horizon/maturity buckets (with and without rollover restrictions)• Basel III standard liquidity measures (Liquidity Coverage Ratio [LCR] and Net Stable Funding Ratio [NSFR]); often

approximated based on assumptions about contractual maturities and credit quality of securities; for LCR, in most cases, the minimum parameters for deposit outflows were chosen; results were checked against the outcome of the preceding quantitative impact study (QIS-6) of the Basel III framework

Risk Horizon One or five working days (one week) and/or one month Risk(s) • Funding liquidity risk: runoff rates, renewal/callback/rollover rates

• Market liquidity risk: valuation haircuts (market-based or predefined) Calibration • Historical experience of banks after the collapse of Lehman Brothers and other episodes of liquidity stresses in the past

• Expert judgment: assumptions about the performance of banks under stress (that is, liabilities runoff, taking into account valuation haircuts to liquid assets, and amortization of outstanding assets)

Other issues • Asset encumbrance• Link to solvency stress test (and scenarios)• Buffer: counterbalancing capacity; offsetting contractual inflows due to central bank support

Benchmarks Metrics/Output • Positive net cash inflow: ability of banks’ liquidity buffers under stressed scenarios to cover expected and potential

outflows over a given time period (that is, liabilities runoff, taking into account valuation haircuts to liquid assets, and amortization of outstanding assets)

• Regulatory liquidity ratio(s): LCR, NSFR, and/or national liquidity risk measure3. Methodology Model • IMF templates and assumptions: (1) implied-cash-flow approach (Čihák 2007; Schmieder and others 2012; Jobst 2017);

(2) LCR/NSFR templates (Catalán 2015)• Regulatory minimum measures: LCR and NSFR (Basel III liquidity risk framework)• Macro-financial model: econometric approach (possibly in combination with solvency feedback effect[s]), for example,

Barnhill and Schumacher 20114. Communication Presentation • Standardized output template for BU and TD results provided to banks and national authorities

• Results discussed in Financial System Stability Assessment (FSSA) (supported by more detailed description of both methodologies and findings in a Technical Note [TN]); in most cases, both FSSA and TN are published

Source: Authors.

tions under stress, in most cases to support national versions of standard liquidity ratios.6 The following sections discuss

6 For instance, the US Federal Reserve Board completes the Comprehen-sive Liquidity Assessment and Review as a complement to the annual Comprehensive Capital Analysis and Review for large financial institu-tions covered by the Large Institution Supervision Coordinating Com-mittee; the scope of the exercises includes 16 firms consisting of US global systemically important banks, US systemically important insur-

ance companies, and international broker dealers with a significant US presence in accordance with the Supervision and Regulation Letter SR 15-7 (April 17, 2015). Similarly, the Hong Kong Monetary Authority conducts the enhanced liquidity stress test, which forms part of the li-quidity reporting framework for banks. The Österreichische National-bank (OeNB) uses a cash- flow- based liquidity stress approach. During the financial crisis in 2008, the Austrian Financial Market Authority and OeNB required banks to report weekly cash flows based on a newly developed standardized liquidity reporting template, which allows the simulation of impact of common shocks based on a uniform methodol-ogy (OeNB 2009; Schmitz and Ittner 2007). See also Appendix 16.7.



September 2010 and December 2016, nine of the 29 FSAPs that incorporated liquidity stress tests (for which detailed information is made publicly available) included nearly all banks in their respective systems (for example, Germany, India, Italy, Korea, Russia, Saudi Arabia, South Africa, Sweden, and the United Kingdom); more than 80 percent of system assets were covered in eight other cases (Austra-lia, Belgium, Canada, China, Denmark, France, India, and Turkey).7

Increasingly, bank liquidity stress tests would also need to be attuned to risks emanating from systemically relevant shadow banking activities and entities (Financial Stability Board 2012; IMF 2014a). As an example, US money market mutual funds are important providers of non- deposit (US dollar) funding to European banks; they were subject to runs themselves during the peak of the crisis in 2008 and had to be rescued either by their bank sponsors or the gov-ernment. Furthermore, banks are sometimes inherently in-tertwined with hedge funds or finance companies (that are dependent on short- term funding), both of which could also be susceptible to funding runs, resulting in spillover effects to the banking sector.

Data

Characteristics Liquidity stress tests require granular bank- level information. The comprehensiveness, comparability, and consistency of these tests depend on the access to quality data that cover essential elements of funding and market li-quidity risks to a sufficient degree of accuracy. Data granu-larity increases with the complexity of the system, including the diversity of sources, and the use of funds:

• TD tests typically rely on confidential prudential in-formation gathered from the supervisory liquidity reporting process. In many cases, the data also cover broad categories of assets and liabilities with break-downs of maturity terms (for example, Australia, Austria, Brazil, Germany, Hong Kong SAR, Korea, Poland, Spain, Sweden, Turkey, and the United Kingdom) and differentiation by currency (for example, Austria, Korea, Singapore, and Turkey). However, in some countries, public data also have been used (for example, Norway and the United States).

• Separately, BU tests using banks’ own data (for example, Belgium, China, Denmark, France, India, Japan, Korea, Singapore, and South Africa) would require careful cross- validation with available super-visory data as well as an assessment of the quality of internal controls, risk management, and corporate

the IMF’s liquidity stress testing framework using examples of applications to FSAPs (Appendix 16.1 provides detailed information alongside these dimensions for 31 jurisdictions). This also includes a brief review of the implementation of li-quidity risk measures under national liquidity reporting frameworks in the context of different types of liquidity stress tests.

Scope

Approach

In FSAPs, surveillance stress testing of banks’ liquidity risk usually consists of either a top- down (TD) approach or, less used to date, a bottom- up (BU) approach. Underlying as-sumptions and calibrations are generally agreed to between the national authorities and the IMF staff:

• TD tests are often conducted by the authorities with inputs from the IMF staff (for example, Austria, Bra-zil, Italy, and Poland) or jointly with the IMF staff (for example, Australia, Belgium, Germany, Hong Kong SAR, Ireland, Russia, Saudi Arabia, Spain, and the United Kingdom) if required owing to the confidential nature of prudential data. However, there are instances where some (or all) TD tests are conducted independently by national authorities (for example, Canada, Hong Kong SAR, India, Korea, Sweden, and Switzerland) or by the IMF staff only (for example, France, Norway, and the United States).

• Increasingly, banks have been involved in BU liquid-ity stress tests for FSAPs (for example, Belgium, China, Denmark, Korea, Singapore, and South Af-rica), which involve both the national authorities and the IMF staff. This approach has enhanced the technical detail of liquidity stress tests, given the granular data available at the bank level compared to the higher- level aggregated information that is used in most TD stress tests.

Coverage

Complete institutional coverage is important for the use-fulness of the exercise. In most financial systems, banks that are systemically important from a solvency perspective also tend to be relevant for the analysis of system- wide li-quidity risk; however, the aggregate effect of many vulner-able, smaller banks with similar business models can also undermine the stability of the system. Sometimes, smaller banks may also account for an important share of liquidity provision in the financial system. In this regard, the selec-tion of relevant banks to include in the sample could be more complicated than in solvency stress tests, where the systemically important institutions may be more obvious. Thus, some FSAP liquidity stress tests covered the entire banking sector, including cooperative and savings banks (for example, Brazil, Denmark, and Switzerland). Between

7 Investment banks (but also foreign branches and subsidiaries), which may not be included in corresponding solvency stress tests, tend to play an important role in funding markets and should ideally be incorporated (for example, Ireland, Hong Kong SAR, and the United Kingdom).



Types of Test Metrics

FSAP stress tests typically assess changes to the funding con-dition of banks under different adverse scenarios within the framework of existing (or useful) liquidity riskmanagement measures. The following policy measures promote a more stable funding profile and improve the resilience of the banks to funding shocks (Nier and others 2014):

• Liquidity buffer requirements encourage banks to hold sufficient liquid assets to cover outflows during time of stress.

• Stable funding requirements ensure that illiquid assets are funded by stable sources of funding.

• Liquidity charges impose a levy on noncore funding.• Reserve requirements ensure that banks hold certain

amounts of reserves with their central bank.• Restrictions on open foreign currency positions and/or

foreign currency- denominated funding aim to limit banks’ exposure to exchange rate risk.

Most exercises combine ICF modeling with standard li-quidity risk measures, benchmarked on national regulatory standards, which are calibrated to (or closely aligned with) the Basel III liquidity framework. Examples include:

• ICF tests: Balance sheet information is used to simu-late a bank- run- type withdrawal of deposits and wholesale funding (including the nonrenewal of con-tracted funding), together with drawdowns of contingent claims and related party funding obliga-tions (usually not decomposed into maturity buck-ets). Cash inflows from contingent funding sources and receipt of payments for maturing claims as well as assumed proceeds from selling available liquid as-sets and/or using them as collateral for secured fund-ing are applied fully, or in part, to counterbalance the assumed outflows. These cash- flow projections may be augmented with market- based measures of the sensitivity of funding costs to changes in the as-set risk of banks based on observed or market- implied default probabilities and expected losses.

• Basel III standard liquidity measures: Under Basel III, banks are expected to maintain a stable funding structure, limit maturity transformation, and hold a sufficient stock of available assets to meet their fund-ing needs in times of stress (BCBS 2010b, 2011, 2012b, 2013a). The framework is based on two stan-dardized ratios, the LCR and the NSFR, which are applied to banks on a consolidated basis.8

• Standard liquidity ratios by national authorities: Many bank regulators have enhanced their national liquidity reporting frameworks to support the imple-mentation of liquidity risk measures (for example, the former UK Financial Supervisory Authority’s li-quidity reporting profile, which has been complemented

governance to include the findings in the FSAP assessment.

• The data cutoff date for a liquidity stress test would ideally coincide with that of the (parallel) solvency stress test, which ensures time consistency in assessing banks’ health and facilitates including feedback ef-fects between the two exercises (if applicable).

Reporting Basis (Consolidation) One dimension of liquid-ity stress that has received little attention so far is the level of consolidation of banks’ financial accounts. Liquidity stress tests may be carried out on consolidated level data (which is very common) or on a legal entity (solo) basis. The latter is only relevant if the stress test includes financial conglomer-ates and/or international groups, which may have consider-able intragroup funding arrangements in place that could be disrupted by cross- border restrictions (that is, “ ring- fencing”) of liquidity (and capital) during times of stress. This aspect becomes even more important in countries where significant market share is held by (1) host- supervised banks, which could experience high liquidity outflows owing to intragroup funding obligations to subsidiaries abroad; or (2) large sub-sidiaries and branches that depend on contingent intragroup funding. For these countries, stress tests have been imple-mented either (1) on a solo basis (for example, Germany, Ire-land, South Africa, and the United Kingdom); or (2) on both solo and consolidated bases (for example, Belgium, Hong Kong SAR, and Singapore). In most FSAPs, however, stress tests have been applied on consolidated data. Cross- border liquidity stress tests using consolidated data are applied in the Spain FSAP IMF 2012b using the Espinosa- Vega and Sole 2011 methodology.

Scenario Design

Once the scope of the liquidity stress test has been deter-mined, the scenario design is defined. It comprises: (1) the definition of the scenarios (for example, scope and severity); (2) the exogenous stress assumptions; and (3) the pass/fail benchmarks. Liquidity stress tests assess the short- term or, in some cases, medium- term resilience of banks to sudden, sizeable withdrawals of funding (liabilities) together with in-sufficient callbacks on outstanding claims. Some tests are aimed at gauging the magnitude of shocks required to cause severe distress, that is, constitute reverse stress tests (“until it breaks”), in addition to “traditional” tests that project li-quidity positions under specified scenarios, usually involving one of the following:

• Cash- flow mismatch analyses over different risk hori-zons, with a focus on the sudden, sizeable withdrawal of short- term funding sources and the sufficiency of selling (unencumbered) existing assets to withstand those shocks under stressed conditions (with asset- specific haircuts).

• Liquidity ratio- based analysis over a longer risk horizon.

8 The Basel liquidity rules only prescribe that the standards be applied on a consolidated basis. Legal entity application is left to national discretion.



recorded either (1) “ off- balance sheet” (for example, liquidity facilities to special investment vehicles and long/short deriva-tives positions) or (2) “ on- balance sheet” if they are “instanta-neous” or have no specific maturities (for example, sight deposits) (Catalán 2015). They receive special treatment, dif-ferent from that accorded to assets and liabilities with non-contingent payoffs and an explicit maturity structure.

Asset Encumbrance The design of liquidity stress tests should, where possible, also include granular information about banks’ asset encumbrance or liquidity from eligible col-lateral ex post haircuts (European Systemic Risk Board 2012). The assessment of banks’ funding risks under stress condi-tions is critically dependent on the market value of liquid as-sets, their current (or expected) encumbrance, and/or the ability to monetize them.

Banks with high asset encumbrance levels (for example, through secured refinancing activities and on- balance sheet structured finance, such as covered bonds) have less capacity to withstand severe liquidity shocks, as their access to collateral- backed funding is constrained. Other unsecured creditors, such as depositors, are also subordinated, increasing the risk of a run during stressful periods. Hence, the liquidity buffer considered in tests comprises only unencumbered liq-uid assets, that is, assets that can be (but have not been) used as collateral to receive funding (except for cash or cash equivalents). The stock of liquid assets normally excludes en-cumbered assets in cases when banks do not have the opera-tional capability to sell them or use them as collateral for a repurchase agreement with the central bank or other banks to meet outflows during the stress period, that is, if repo op-erations for commercial and/or central bank money are not possible.

In most cases, liquidity reporting requirements of banks already include assumptions on asset encumbrance affecting the valuation of liquidity buffers and/or assumptions on the depletion of funding sources under stress (“behavioral ad-justments”), such as in the United Kingdom. In other cases (for example, Australia, Belgium, Brazil, Hong Kong SAR, Italy, Japan, Korea, and Turkey), any encumbered assets were excluded from the scope of liquid assets from the exer-cise. Stress tests that include public data or supervisory data, for which a consistent application of these adjustments can-not be verified, assume a uniform degree of asset encum-brance for the valuation of liquid assets, in addition to the application of haircuts (Jobst 2017).

Link to Solvency Estimated changes in funding costs dur-ing times of stress (and their impact on net cash flows) can help link liquidity scenarios to the capital adequacy assess-ment in solvency stress tests. The macro- financial transmis-sion of shocks affecting the capital assessment also applies to corresponding liquidity stress tests insofar as any change in funding costs affects the assumptions of interest expenses (and cost of capital) applied to the solvency tests. Solvency stress tests in recent FSAPs estimate the impact of shocks to banks’ balance sheets through the cost of funding of

by the liquidity metric monitor,9 and the National Bank of Belgium’s liquidity ratio). Most standard li-quidity ratios are assessed as noncumulative mea-sures of potential liquidity shortfall for stress periods covering the short- and medium- term resilience of individual banks and the overall system.

Risk(s) and Risk Horizon These tests cover both funding and market liquidity risks. In most FSAPs, stress tests are mod-eled as cash- flow tests of bank- run- type funding shocks over short consecutive periods. Liquidity metrics focusing on structural asset- liability mismatches (similar to the NSFR) are applied to longer horizons. Commonly, the former is used to analyze either consecutive (cumulative) daily cash outflows over several days (typically five working days or one week) or one- off, noncumulative aggregate cash outflows over 30 days, whereas the latter assesses the adequacy of stable sources to continuously fund cash- flow obligations in-side a one- year time horizon.

Calibration Most liquidity stress tests in FSAPs entail deter-ministic stress scenarios based on ICF approaches and fully fledged cash- flow tests. This is distinct from simulation ap-proaches (possibly combined with network modeling), which have also been used in past exercises, albeit less fre-quently. Scenario assumptions are meant to be sufficiently “extreme yet plausible” to effectively cover the scope of exist-ing vulnerabilities. This application is particularly challeng-ing during benign times when there is greater uncertainty about the realization of risks. In some cases, it might also be useful to reconcile the characteristics of liquidity shocks with the sudden- stop and boom- bust scenarios of solvency stress tests. ICF tests and standard liquidity measures, in-cluding regulatory ratios such as LCR and NSFR, contain a predefined set of assumptions (which can be subject to sensi-tivity analysis). Other deterministic stress tests may be based on historical worst- case scenarios, expert judgment, or statis-tical models/valuation approaches (and then mainly on the asset side).

Other Considerations

The scenarios define the scope of the liquidity buffer as well as the contractual maturities of expected cash flows in stress situations. The quantification of assets and liabilities generat-ing cash flows should, if possible, be supplemented with as-sumptions about potential cash flows from related and third parties in the form of committed but unused credit lines/ liquidity facilities. These contingent claims/liabilities are an essential element of projected behavioral cash flows and are

9 The liquidity metric monitor is designed to demonstrate some of the li-quidity metrics calculated by the Prudential Regulation Authority us-ing prudential information in accordance with FSA047 and FSA048. It also provides estimates of the Basel III liquidity ratios (LCR and NSFR). See https://www.bankofengland.co.uk/ prudential-regulation /publication/2013/supervisory-tools-liquidity-tools.




minimum reserve requirements) as well as unencumbered assets, which could generate inflows from outright sales or collateralized lending (“secured funding,” for example, repo and securities lending transactions). The evolution of trading assets in response to market risk shocks, such as to foreign exchange rates and interest rates, determines the degree of illiquidity affecting both price/valuation changes of fixed in-come holdings and their speed of disposal.

The buffer may be applied to cover short- term payment obligations (that is, available assets that could be sold under stress). Different haircuts are imposed on assets that are in-cluded in the buffer, depending on their perceived (or as-sumed) liquidity under stress. Haircuts account for estimated valuation losses due to potential illiquidity of tradable expo-sures and the resulting changes in funding (costs).11 Recent European FSAPs (for example, Belgium, France, Germany, Italy, Spain, and the United Kingdom) have also acknowl-edged sovereign risk by estimating haircuts for relevant gov-ernment debt holdings based on the impact of changes to credit risk on bond prices assuming an increase in sovereign default risk that is consistent with market expectations im-pacting the valuation of local and foreign government debt (see Chapter 9).

Contractual and behavioral cash inflows. The amortization of existing (contractual) claims/obligations, depending on the renewal/ call back rate, and the emergence of contingent (behavioral) assets/liabilities generate cash flows, which can be modeled on a cumulative (that is, multiperiod) or noncu-mulative basis. Contractual cash flows remain firm and un-changed under stress while behavioral cash flows could change significantly.12 Behavioral flows could either mitigate or amplify contractual cash flows through (1) additional in-flows related to either new secured and unsecured funding at shorter but also longer maturity terms (for example, as new deposits, wholesale funding, and debt issuance) or rollover/refinancing of contractual liabilities (for example, part of the maturing time deposits are likely to be rolled over); and (2) additional outflows associated with expected new loans, investments, or undrawn committed credit lines.13

short- term debt and the maturing portion of long- term debt with a lag (for example, Brazil, Germany, Spain, and the United Kingdom).

Changes in funding costs influence the expected availability and maturity tenor of available funding over the risk horizon. Banks’ applications of internal pricing mechanisms, which of-ten include hedging of funding cost changes, are also important elements of the chosen cost- of- funding method. These costs typically take the form of an additional interest expense. The elasticity of funding costs is nonlinear to changes in solvency conditions and could be differentiated across maturity tenors and types of funding, such as checking/term deposits, secured/unsecured wholesale funding, and short- term debt that would need to be rolled over within the risk horizon of the stress test. In the case of noncommoditized bank debt, such as interbank funding arrangements, lending rates adjust in response to changes in counterparty risk.

Liquidity stress tests should, where possible, incorporate feedback (or second- round) effects when considering the re-action to funding shocks. Funding costs are influenced by banks’ solvency conditions and changes in market prices during stress periods.10 Impairment losses could also raise funding costs (Aiyar and others 2015) in a dynamic between bank liquidity and solvency, but this issue remains to be addressed in stress tests. Outright rationing of funding, in addition to increases in cost, may arise for banks that are perceived to be weak vis- à- vis their peers. Moreover, liquid-ity stress can spill over to other (stronger) banks by affecting market liquidity and, ultimately, the availability of funding for these banks, which could lead to solvency concerns. In this regard, sources of macro- financial shocks can be trig-gered, or at least propagated, by vulnerabilities to the adverse effects of such interactions in times of collective distress. Finally, there can be additional spillover effects associated with counterparty risk if weak banks are unable to honor, in part or entirely, their interbank exposures. However, the operational implementation of feedback loops in the context of system- wide stress tests remains at a seminal stage (Appendix 16.4).

Liquidity Buffer Counterbalancing capacity. The liquidity buffer represents

banks’ “counterbalancing capacity” under stress. It com-prises cash and cash balances with central banks (excluding

10 Schmitz, Sigmund, and Valderrama 2017 find evidence of nonlinear ef-fects between solvency and funding costs using a simultaneous- equation approach drawing on supervisory data for 54 large banks from six ad-vanced countries between 2004 and 2013. The study confirms earlier evi-dence in Annaert and others 2013, which shows that the interaction between solvency and funding costs is indeed significant in a sample of 31 large euro area banks over the precrisis period from 2004 to October 2008. Similarly, Hasan, Liu, and Zhang 2016 show that solvency has significant impact on bank funding costs, using a sample of 161 global banks from 23 countries between 2001 and 2011. This is confirmed by Caceres and others (2016) when they examine the sensitivity of bank funding costs to bank solvency drawing on the Federal Deposit Insurance Corporation call re-port covering 10,000 US banks between 1993 and 2013.

11 Haircuts would ideally be applied irrespective of whether assets are held in the trading or banking books (since a bank’s access to funding mar-kets [and thus its funding costs] will depend on the market’s current valuation of the bank’s entire portfolio and not on the accounting valu-ation on a hold- to- maturity basis). See Chapter 9 for a discussion of the scope of valuation haircuts in the context of sovereign exposures.

12 Projected cash flows that stem from contractual rights or obligations and have a known maturity date are differentiated from those that are likely to materialize but have not yet been contracted and could exceed expectations (based on historical experience) or existing cash reserves.

13 Note that expected cash inflows (outflows) reflect changes in available (required) funding through assets (liabilities); however, this relation-ship reverses for potential cash flows. For instance, the possible use of a committed credit line (as an asset) by a related or third party to obtain funding during stress represents a potential cash outflow, while access to contingent intragroup funding (as a liability) contributes to potential cash inflows.



Usually, one or more quantitative stress test methods are applied to estimate potential liquidity shortfalls under the predefined shocks. Peer comparisons facilitate the assess-ment of liquidity risk relative to other banks within and out-side the banking sector, based on common scenarios and benchmarks, especially if assumptions carry a high degree of uncertainty. For example:

• TD tests are commonly configured as ICF tests, which complement standard liquidity ratios and minimum prudential requirements. For banks with simple funding structures, these tests are the most appropriate. A key prerequisite for carrying out ICF tests is access to a wide range of data on contractual cash flows for different maturity buckets and possi-bly behavioral data based on banks’ financial/fund-ing plans. In addition to assessments of maturity mismatches for specific time horizons under stress, they also include duration gap analyses. Determinis-tic liquidity stress tests developed by Čihák (2007), Schmieder and others (2012) as well as Jobst (2017) are applied in most FSAPs (see Appendix 16.5).15

• Regulatory minimum measures such as the LCR and NSFR from the Basel III liquidity risk frame-work have become staple stress test methods.

• For more sophisticated financial systems (and banks) for which market data are available, stochastic meth-ods may be used as a complement. These market- based models incorporate uncertainty using historical volatility and/or market information (for example, Jobst 2011 and 2012). They allow for sensitivity anal-ysis (that is, stress of one risk factor/type) or scenario analysis (that is, stress of multiple risk factors/types).

• Macro- financial econometric models that combine liquidity stress with solvency feedback effects have been developed but their application remains scarce (for example, Barnhill and Schumacher 2011; Valder-rama 2016; Gray and others 2017; Krznar and Matheson 2017).

Communication

Presentation

The main objective of stress tests is to draw the attention of bank management, supervisors, and regulators to potential risks and, if necessary, to galvanize action aimed at mitigat-ing the impact of associated vulnerabilities. As noted in Jobst, Ong, and Schmieder 2013, it is important that the findings be appropriately conveyed. In FSAPs, liquidity stress tests are based on bank- by- bank analyses but results are generally aggregated for confidentiality reasons, under-scoring the importance of meaningful firm- level data. Hence, the templates that are designed by the FSAP team for

The most important funding sources that generate cash inflows are:

• Contractual: Expected cash inflows related to (1) the repayment of amortized lending with/without liquid financial assets as collateral (that is, secured/ unsecured lending); and (2) transactions with liquid securities and bank loans (that is, asset sales) and funding from related parties (intragroup funding).

• Behavioral: Potential cash inflows from related and third parties in the form of committed or uncom-mitted but unused credit lines/liquidity facilities as contingent liabilities (situation on reporting date).

Scenarios encompass both systematic shocks affecting all banks and idiosyncratic shocks that impact individual banks only. Given that marketwide stresses amplify the individual liquidity risk, Schmieder and others 2012 advocate includ-ing combined scenarios similar to the one underlying the LCR. Where possible, scenarios should also be accompa-nied by consistent narratives underpinning the assumptions on all relevant cash- flow parameters and risk factors, includ-ing: (1) callback rates for lending and runoff rates for fund-ing,14 (2) valuation haircuts for assets sold at fire sale prices and drawings of contingent liabilities (Coval and Stafford 2007; Shleifer and Vishny 2010), and (3) the impact of banks’ rating downgrades as a result of deteriorating sol-vency conditions.

Benchmark(s)

The existing framework caters largely to TD stress tests. Hence, the emphasis is placed on running a set of consistent tests for all banks within a system (and relevant banks and nonbanks outside of it, if necessary) against common bench-marks, such as positive net cash flows, the ability of banks’ liquidity buffers to withstand stressed scenarios, and regula-tory liquidity ratios.

Methodology

The methods selected for FSAP liquidity tests depend on the sophistication of the relevant banking sector. Considerations include, among others:

• The importance of deposits relative to wholesale- based funding

• The role of off- balance sheet funding (especially via structured finance and derivatives transactions)

• The concentration of lending to related parties in the banking book

• The nature of counterparty risk (for example, the rel-evance of market- based transmission channels of funding impacting the availability and pricing of funding, such as margin calls)

14 This also includes the discount factors for contingent claims and liabili-ties to related and third parties.

15 A comprehensive ICF test approach and the related tool have been ap-plied in FSAPs (Jobst 2017), which can be found on the IMF eLibrary.



liquidity stress test component of their respective FSAPs (that is, Argentina, Austria, Belgium, Brazil, Canada, Den-mark, France, Germany, Hong Kong SAR, Ireland, Italy, Korea, Norway, Russia, South Africa, Sweden, Switzerland, the United Kingdom, and the United States). For the re-maining countries, the basic information on the methodolo-gies and results of the stress testing exercise are published in the main FSAP document, the FSSA report.

4. CAVEATSLiquidity stress tests that aggregate individual liquidity risk measures across all banks do not necessarily capture the scope of system- wide risks. Prudential measures, such as ICF tests, or standard indicators, such as prudential ratios, have an institutional focus. They assume that sufficient liquidity greatly reduces the likelihood of firm- specific funding short-falls and any associated knock- on effects on capital ade-quacy. So, they ignore the system- wide effects from potential herding behavior by banks and their joint sensitivities to funding shortfalls.

Liquidity stress tests do not explicitly assume potential refinancing via central banks that act as lenders of last resort. Assumptions made on haircuts to unencumbered assets for banks’ counterbalancing capacity are often silent as to whether banks need to seek central bank funding or may still be able to secure wholesale funding (which is likely to be severely impaired during times of stress). Sufficient liquidity in interbank markets implies that central banks would only be required to act as lenders of last— not first— resort (Jobst 2014). However, funding shocks often represent extreme outcomes, which could lead banks to draw on their “expen-sive” liquidity buffers to cover the probability of tail events.

Liquidity stress test results need to be put in context given their static nature and the implicit assumption that all banks face escalating liquidity risk at the same time. Depending on the stress testing methodology, any estimated liquidity shortfall is assumed to be the result of coincidental and mostly predefined funding shocks. The results should be in-terpreted in terms of a general vulnerability to certain as-sumptions characterizing adverse liquidity conditions, rather than being representative of actual liquidity needs (given the role played by central banks as a liquidity backstop). In other words, the calculated effect might overstate the actual im-pact from the realization of assumptions about varying cash- flow scenarios. In addition, Schuermann 2012 cautions that the “dynamism” of liquidity positions that are subject to rapid change means that stress test results may not be very informative by the time they are disclosed.

Stress test results need to be suitably qualified based on mitigating considerations. An example would be the likely reallocation of deposits within the banking sector when funding shocks do not affect all banks simultaneously; in most cases, deposits largely remain in the banking sector and are swiftly reallocated with weaker banks, which would need to offer above- average deposit rates to retain or attract

input by the authorities (Appendix 16.6, Appendix Fig-ure 16.6.1) are:

• consistent with any local regulatory requirements and, where relevant, any international regulatory standards (for example, Basel III) for cross- country comparison purposes; and

• sufficiently granular, showing (1) peer groups; (2) some measure of dispersion, such as different “buckets” of liquidity ratio results or maturity tenors; (3) the number of banks failing to meet the liquidity bench-mark; (4) the percentage of total sample assets of banks failing to meet the benchmark; and (5) detailed assumptions, which also clarify key limitations to the implementation of the stress test.

As with solvency stress tests, the findings of liquidity stress tests are used for two main purposes: (1) to provide quantita-tive support for FSAP stability risk assessments by estimating the impact from the realization of the predefined shocks, and (2) to facilitate policy discussions with the authorities on risk- mitigation strategies and crisis preparedness.

Publication

The communication of stress test results is a critical element of any publicly announced stress testing exercise, especially if enhanced transparency has immediate benefits for financial stability. Any published analysis should aim to provide a complete assessment of the system- wide resilience to liquidity risk while avoiding causing either complacency or undue alarm. Moreover, the disclosure of system- wide liquidity con-ditions (if based on prudential data) is particularly sensitive given that market participants may be able to take positions against those banks in short- term money markets. For FSAP exercises, the following aspects are especially relevant:

• The objectives, definitions, assumptions, methods, and limitations of stress tests are provided in Technical Notes on Stress Testing and/or, as supplementary infor-mation, in FSSA reports. Publication of these docu-ments is voluntary for country authorities. The presentation of the stress test results requires the use of a standard format, that is, the STeM, to improve trans-parency and facilitate cross- country comparisons.

• The aggregated results are almost always disclosed. As a minimum, information about the performance of banks under stress (that is, liabilities runoff, after accounting for valuation haircuts to liquid assets and the amortization of outstanding assets) is presented in the form of liquidity ratios and/or maximum days of resilience. As with solvency stress tests, authorities rarely agree to make available the liquidity stress test results of individual banks.

To date, all the jurisdictions in this chapter’s sample (30) have authorized the publication of at least the main results and general information on the stress testing framework of the FSAP. Almost two thirds (19) have authorized the full publication of Technical Notes containing all details of the



but there are general principles and elements that are com-mon to all jurisdictions and facilitate the consistent implementation of comparable liquidity stress tests.

As with all other aspects of stress testing, the evolving nature of bank business models, financial instruments, and capital market conditions requires adaptability. The future of liquidity stress testing will likely be multipronged with a shift toward comprehensive cash-flow- based tests. Liquidity stress testing approaches will also require a deeper under-standing of the interrelationship between solvency and liquidity risks.16

depositors. Other mitigating factors include: (1) contractual capital inflows from maturing wholesale lending, (2) possible central bank support via committed liquidity facilities and widening of eligible collateral, and (3) the likely compensat-ing outcome for the system from the deposit insurance scheme.

5. CONCLUSIONThis chapter provides a conceptual overview of liquidity stress testing of banks in the context of the IMF staff’s application in FSAPs for countries with systemically impor-tant financial sectors. The implementation of these stress tests varies— depending on the structural characteristics of financial systems and national prudential requirements—

16 In this regard, the work of the Research Task Force of the Basel Com-mittee on Banking Supervision on liquidity stress testing provides useful insights into the important interaction of liquidity and solvency risks.


Appendix 16.1.FSAP Liquidity Stress Tests since

FY2011


Macroprudential Liquidity Stress Testing in FSA



418


Liquidity Stress Test Matrix (STeM) for FSAPs of Systemically Important Financial Systems (Illustrative)

1 G20

Indonesia

3 S29

Netherlands

4 S29

Luxembourg

6 S29

Sweden

9 G20, S29

China

10 G20, S29 Mexico

11 G20

Saudi Arabia

12 S29

Spain

13 G7, G20, S29

Japan

15 G20, S29 Australia

16 G20, S29

Brazil

Publication DateStress Testing Framework

September 2010 June 2011 June 2011 July 2011 November 2011 March 2012 April 2012 June 2012 August 2012 November 2012 June 2012

1. Scope

Approach • TD by authorities (Bank Indonesia), in collaboration with IMF staff.

• TD by authorities (De Nederlandsche Bank).

• TD by authorities (Riksbank). • BU by banks. • TD jointly by authorities (Saudi Arabia Monetary Authority) and IMF staff.

• TD by authorities (Banco de España), in collaboration with IMF staff.

• BU by banks using supervisory templates and assumptions.

• TD by authorities (Bank of Japan).

• TD by authorities (Australian Prudential Regulation Authority) using IMF templates and assumptions.

• TD by authorities (Banco Central do Brasil), aligned with IMF framework (used as a benchmark).

Coverage

Institutions • All (121) banks, except rural banks, which account for about 1 percent of total financial sector assets, and are subject to different supervisory standards and regulatory requirements.

• 11 largest banks. • 4 largest banks (plus group of large European banks as a benchmark).

• 17 largest commercial banks.

• 12 locally incorporated banks.

• 29 banks. • BU: Three mega banks.• TD: All major and regional banks (111),

including 11 major banks, 63 “tier 1” regional banks, and 37 “tier 2” regional banks.

• 5 largest banks. • All banks (137).

Market share • 99 percent of total sector assets.

• >90 percent of total sector assets.

• 90 percent of total sector assets.

• 83/66 percent of total assets of commercial banks/all banks.

• 98 percent of total sector assets (large banks: 59 percent; medium-sized banks: 26 percent; small banks: 13 percent).

• 91 percent of total sector assets. • BU: About 50 percent of the banking sector (40/30 percent of total sector assets based on a narrow/broad definition of the banking sector).6

• TD: About 77/62 percent of total sector assets based on a narrow/broad definition of the banking sector.6

• 80 percent of total sector assets. • 100 percent of total sector assets.

Data

Source • Supervisory data. • Supervisory data. • Supervisory data. • Banks’ own data. • Supervisory data. • Supervisory data. • BU: Banks’ own data.• TD: Supervisory data.

• Supervisory and banks’ own data.

• Supervisory data.

Cut-off date • End-Q3 2009. • End-Q2 2010. • End-Q3 2010. • End-2010. • End-2010. • End-2011. • End-Q3 2011. • End-Q1 2012. • End-2011.

Reporting basis • n.a. • Consolidated banking groups.

• Consolidated banking groups. • Consolidated banking groups.

• Consolidated local entities.

• Consolidated for domestic business only.

• Consolidated banking groups. • Consolidated banking groups. • Consolidated banking groups.

2. Scenario Design

Test(s) • Implied cash flow test calibrated to a short-lived episode of liquidity stress experienced between September and October 2008, assuming a sizeable withdrawal of deposit funding and restricted funding access to interbank funding markets, subject to a deteriorating counterbalancing capacity via the sale of of existing assets to withstand those shocks under stressed conditions.

• Implied cash flow test over 12 months, assuming deposit run, dry-up of wholesale funding markets, and haircuts on liquid assets (as counterbalancing capacity). • Buffer: Counterbalancing capacity.

• Liquidity measures similar to the Basel III liquidity ratios (LCR and NSFR): on the assets side, cash is allocated a zero weight while loans are allocated an 85 percent weight; on the liabilities side, equity capital and liabilities maturing in less than a year are allocated a weight of 100 percent, while short-term market funding is allocated a zero weight; unexpected cash outflows from sudden, sizeable loss of funding (wholesale/deposits), drawdown of unused credit lines, and the sufficiency of existing assets to compensate for those shocks under stressed conditions.

• Tests combine funding liquidity with haircut on liquid assets.

• Three scenarios for each risk horizon: (i) 7-day: decline in bond prices (1, 3, and 5 percent), deposit drawdown (2, 4, and 6 percent), decline in interbank funding (5, 10, and 15 percent), and higher required reserve ratio (0, 0.5, and 1 pcp); (ii) 30-day: share of maturing loans that become NPLs (4, 7, 10 percent), decline in bond prices drop (3, 5, and 8 percent); deposit drawdown (4, 6, and 8 percent); decline in interbank funding (5, 10, and 15 percent), and higher required reserve ratio (0, 0.5, and 1 pcp).

• General deposit run of 25 percent.

• Short-term deposit run of 40 percent.

• Implied cash flow test (bank-run type funding shock for 5 and 30 consecutive days).

• LCR, NSFR, maturity mismatch analysis.

• BU: Cash flow mismatch analysis (implied cash flow test) over two different risk horizon (1 week and 1 month), with focus on the sudden, sizeable withdrawal/market freeze of U.S. dollar funding (with run-rates applied to interbank funding and deposits by banks/central banks) and the sufficiency of selling (unencumbered) existing assets to withstand those shocks (with asset-specific haircuts); U.S. dollar funding through FX spot purchases and BoJ’s U.S. dollar funds supplying operations are excluded as mitigating actions.

• TD: Liquidity ratio-based analysis based on the impact of a withdrawal/market freeze of wholesale funding and deposit withdrawal over a 3-month risk horizon; no haircut applied to the liquidity value of government bonds (JGB), which constitutes a large share of liquid assets.

• Buffer: Counterbalancing capacity taking into account haircuts to liquid assets.

• Cash flow mismatch analysis: Implied cash flow test of bank-run type funding shock for 5 consecutive periods (30 days), with focus on the sudden, sizeable withdrawal of funding and the sufficiency of selling (unencumbered) existing assets to withstand those shocks under stressed conditions (with asset-specific haircuts); only but only on-balance sheet items (i.e., no contingent claims/liabilities); also maturity mismatch analysis.

• Buffer: No consideration of the RBA Committed Liquidity Faciliy (which is permitted under LCR).

• Cash flow mismatch analysis based on bank-run type funding shock developed by authorities (similar to LCR): Run-off rates for funding and market liquidity stress levels (for unencumbered assets) equivalent to historical evidence at the 99th percentile, simulated for a 21-day stress period, and without recourse to the reserve requirements. For the run-off rates, the bank-specific concentration of funding and the historical volatility of deposits were taken into account.

• Basel III ratios (LCR and NSFR, based on “old” 2010 BCBS rules) for 15 banks.


Benchmarks

Metrics/Output • Performance of banks under stress (i.e., liabilities run-off, taking into account valuation haircuts to liquid assets, and amortization of outstanding assets).

• Number of months banks can withstand stocks, accounting for the concentration of liquid assets.

• Ability of banks’ liquidity buffers under stressed scenarios to cover unexpected outflows over three months.

• Relates the weighted average of liabilities to the weighted average of assets.

• Performance of banks under stress (are they able to maintain a liquidity ratio of 25 percent and a liquidity gap of zero).

• Performance of banks under stress (are they able to maintain regulatory liquidity ratio, i.e., liquid assets to deposit liabilities, of 20 percent).

• Performance of banks under stress (how long they can withstand shock, how many banks fail, liquidity needs, Basel III ratios, etc.).

• BU: Performance of banks under stress (i.e., deposit withdrawal, liabilities run-off [interbank funding], and cash outflows of commited asset-backed commercial paper and credit facilities), taking into account valuation haircuts to liquid assets) under two scenarios (adverse/extreme). Hurdle metrics: Net cash inflow position after mitigating actions.

• TD: Performance of banks under stress (i.e., deposit withdrawal and liabilities run-off [interbank funding]) under two scenarios: (i) market freeze scenario (no wholesale funding but also no deposit run-off), and (ii) deposit withdrawal scenario (escalates market freeze scenario by assuming deposit run-off by 5-10 percent. Hurdle metrics: Liquid asset ratio (> 100 percent).

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets but only on balance sheet items).

• Hurdle metrics: Asset and liability maturity profile, asset-liability gap (funding shortfall).

• Performance of banks under stress: pass rate based on hurdle rate for all three ratios (the BCB liquidity ratio, the LCR and the NSFR).

• Analysis of key drivers by banking group (large/medium-sized/small banks, foreign banks): Comparison of outflow of funds and inflow of fire sale assets relative to total assets and distribution of corresponding liquidity ratios under stress; relative performance of banks with respect to LCR and NSFR (to identify banks vulnerable in both dimensions); sensitivity analysis for run-off rates.

3. Methodology

Model • B/S using authorities’ templates and assumptions.

• B/S using authorities’ templates and assumptions.

• B/S using authorities’ templates and assumptions.

• B/S. • B/S. • B/S using IMF templates and assumptions.

• B/S. • B/S. • B/S.

4. Communication

Publication • No Technical Note.• Methodology and

results published in FSSA.

• No Technical Note.• Methodology and

results published in FSSA.

• Technical Note, published.• Results discussed in FSSA,

published.

• Technical Note, not published.

• Results discussed in FSSA, published.



• No Technical Note.• Methodology and results

published in FSSA.

• No Technical Note.• Methodology and results published in

FSSA.

• Technical Note, not published.• Results discussed in FSSA;

technical details included as Appendix.

• Technical Note, published.• Results discussed in FSSA, published.

URL https://www.imf.org/external/pubs/ft/scr/2010/cr10288.pdf

https://www.imf.org/external/pubs/ft/scr/2011/cr11144.pdf

http://www.imf.org/external/pubs/ft/scr/2011/cr11288.pdf







Contributors:

Stress testing team* J. Gobat R. Vermeulen

Central bank routine stress test results shared with FSAP team.

N. Oulidi S. Stolz

M. Cihak E. Yèhoué A. Jobst L. L. Ong S. Kwoh

S. Arslanalp I. Kaminska (INS) C. Schmieder





http://www.imf.org/external/pubs/ft/scr/2011/cr11321.pd






Andreas A


hristian Schmieder

419(continued)

17 G7, G20, S29

France

18 G20, S29

India

19 G20

European Union

20 S29

Belgium

21 S29

Poland

22 G7, G20, S29

Italy

23 S29

Singapore

24 S29

Austria

25 G7, G20, S29

Canada

26 S29

Hong Kong SAR

27 S29

Switzerland

28 S29

Denmark

December 2012 January 2013 March 2013 May 2013 July 2013 September 2013 November 2013 January 2014 February 2014 July 2014 May 2014 December 2014

• BU by banks using IMF staff/Banque de France templates and assumptions following EBA liquidity risk assessment.

• TD by IMF staff.

• BU by banks using joint IMF staff/Reserve Bank of India templates and assumptions.

• TD jointly by authorities and IMF staff.

• BU by banks using authorities’ (National Bank of Belgium) templates and assumptions.

• TD by authorities, in collaboration with IMF staff.

• TD by authorities (National Bank of Poland) with inputs from IMF staff.

• TD by authorities (Banca d’Italia) with inputs from IMF staff.


• TD by authorities (Monetary Authority of Singapore).

• TD by authorities (Oesterreichische Nationalbank) with inputs from IMF staff.

• TD by authorities (Bank of Canada and Office of the Superintendent of Financial Institutions).

• TD by authorities (Hong Kong Monetary Authority), including Wong-Hui (2009) approach.

• TD by IMF staff (with data input from authorities), using IMF templates and assumptions.

• TD by authorities (Swiss Financial Market Supervisory Authority and Swiss National Bank).

• TD by IMF staff.

• BU by banks using supervisory (Finanstilsynet) templates and assumptions.

• 8 largest banks. • BU: 10 commercial banks.• TD: 60 scheduled

commercial banks.

• 6 largest banks (and entire banking sector excluding foreign branches, Euroclear and Bank of NY Mellon [for NBB Liquidity Ratios]).

• 20 largest banks. • 33 largest banks, including 6 foreign banks’ subsidiaries.3

• 3 domestic banks, 1 foreign subsidiary, and 3 foreign branches.

• 29 banks (subject to weekly cash flow liquidity reporting).

• 6 commercial banks. • TD: All large, locally incorporated, licensed banks (19) and selected foreign branches (8).

• Wong-Hui (2009) approach: Largest locally incorporated, licensed banks (9) and two foreign branches.

• TD by authorities: Almost all banks.

• TD by IMF staff: 30 banks.

• Excess liquidity coverage test: 81 banks.

• LCR: 16 banks and mortgage credit institutions (MCIs).

• Funding ratio: 85 banks.• 80 percent of total

sector assets. • BU: 50 percent of total sector assets.

• TD: 99 percent of total sector assets.

• BU: 86 percent of total sector assets (excluding foreign branches).

• TD (Basel III measure): 82 percent of total sector assets (excluding foreign branches) on a solo basis and 90 percent on a consolidated basis.

• TD (NBB Liquidity Ratios): 93 percent of total sector assets (excluding foreign branches).


• >90 percent of total sector assets. • 74 percent of total sector assets.



• TD by authorities: 68 percent of total sector assets.

• TD by IMF staff: 60 percent of total sector assets (Basel III standard measures).

• Wong-Hui (2009) approach: 63 percent of total sector assets.

• TD by authorities: Almost total sector assets [TD authorities].

• TD by IMF staff: 85 percent of total sector assets.


• BU: Banks own data.• TD: Publicly available

data (results not reported).

• BU: Banks’ own (proprietary) data .

• TD: Supervisory data.

• BU: Banks’ own (proprietary) data.

• TD: Supervisory data.

• Supervisory data. • Supervisory data (including data on sovereign risk, collateral, and retail deposit volatility in weekly/monthly intervals).

• Banks’ own data. • Supervisory data. • Banks’ own data. • Supervisory data. • Authorities: Supervisory data.• FSAP team: Publicly available

data.

• Supervisory and banks’ own data.

• End-2011. • End-Q2 2011. • End-Q2 2012. • End 2012. • End-2012 (for liquidity position data).

• End-March 2013 (for rating and other market valuation data).

• End-Q1 2013. • End-2012. • End-April 2013. • End-Q2 2013. • End-2012. • End-Q1 2014.

• Consolidated banking groups.

• Unconsolidated domestic businesses.

• BU: Solo basis.• TD: Solo/consolidated basis.• Only unencumbered liquid

assets (generating cash inflows), ie, that can be sold or used as a collateral to receive funding (with the exception of cash/cash-equivalents).

• Solo basis for largest 20 commercial banks (including both domestically controlled banks and subsidiaries of foreign banks).

• Consolidated banking group for domestic banks (based on Bank of Italy’s standard weekly liquidity monitoring data), covering short-/medium-/long-term maturities for both retail deposit and wholesale funding, including durations.

• Consolidated banking group for domestic banks.

• Solo basis for foreign subsidiaries and branches.

• Consolidated banking group for domestic banks; granular data based on contractual and behavioral expected cash flows over five maturity buckets (5 days, 1 month, 3 months, 6 months, and 12 months).

• Consolidated banking group. • Solo basis, with the exception of consolidated/combined basis for selected locally incorporated, licensed banks (10) for Basel III standard measures (LCR and NSFR).

• Consolidated basis for Wong-Hui (2009) approach.

• n.a. • Unconsolidated (for excess liquidity coverage test and funding ratio).

• Consolidated (for LCR).

• Systemic and idiosyncratic risk.

• Bank run and drying up of wholesale funding markets, taking into account haircuts to liquid assets.

• A 30-day deposit and, separately, wholesale funding withdrawal of 10 percent.

• A 5-day deposit (wholesale funding) withdrawal of 5 (3) percent.

• Maturity mismatch.• Rollover risk.

• BU: LCR and NSFR (old version, Dec. 2010) [BU].

• TD: LCR (revised as per guidance published in Jan. 2013 [including assessment of haircuts on liquid assets, assumption on expected and contingent cash in- and outflows]) and NBB Liquidity Ratios (one week/one month) reflecting a bank-run type market/funding liquidity risk scenario similar to the LCR.

• Also alternative scenarios for NBB Liquidity Ratios: (i) the absence of a deposit run; (ii) the escalation of sovereign risk (requiring higher valuation haircuts for collateralized funding with central banks); and (iii) the absence of contingent cash inflows from related parties.

• Only unencumbered liquid assets (generating cash inflows), ie, that can be sold or used as a collateral to receive funding (with the exception of cash/cash-equivalents).

• Liquidity shock scenarios defined by authorities.

• Cash flow mismatch analysis over 30-day horizon with cash outlows due to refinancing risks, with wholesale funding and deposit outflows/reduction of liquidity buffer due to sovereign and bank downgrades (increasing ECB haircuts) and declining market valuation of sovereign debt securities.

• Scenarios: (i) “Adverse scenario” (motivated by actual distress experience at end-2011), including refinancing risk (0 percent rollover), changes to ECB haircut (up to two-noth downgrades), increased volatility of deposits (up to 33 percent depending on type, with LCR-prescribed outflow rates as floors), widening of credit spreads; and (ii) “Alternative scenario” (focusing on market factors), which excludes deposit outflows.

• Buffer: Counterbalancing capacity through unencumbered securities eligibile as collateral for ECB refinancing, assessed at market values tnet of ECB haircuts (at security-by-security level).

• BU: Cash flow mismatch analysis.

• TD: Basel III framework (LCR as per revised guidance published in Jan. 2013).

• Buffer: Minimum liquid assets reserve [BU] and monetary authority’s liquidity facilities.

• Cash flow mismatch analysis using six major currency baskets, with scenarios based on macro scenario of the solvency stress test: (i) PD shifts feed into the counterbalancing capacity and cash inflows; (ii) feeback effects are included due to the rising funding costs projected under the adverse macro scenario.

• Scenarios are grouped into a baseline, “market mild,” “market medium,” “market severe,” and combined scenario (including market and idiosyncractic shocks), including instantaneous outflow of funding and gradual outflow over 30-day, 90-day, and 1-year horizon.


• MFRAF: Supervisory model to quantify the funding liquidity risk (and network effects) as endogenous outcome of the interaction between market liquidity risk, solvency risk, and the structure of the sample banks’ funding under different scenarios; approach mimicks the effect of a noncumulative liabilities run-off based on a sudden, sizeable withdrawal of funding (short-term liabilities) over a six-month time horizon, triggered by a certain level of credit losses, and the sufficiency of existing assets to withstand this shock. Changes in liquidity are translated into losses affecting the capital position of each bank; calibrated parameters of the liquidity measure are in large part consistent with parameters of the LCR.

• Buffer: Minimum liquid assets reserve.

• TD by authorities: HKMA’s 7-day test (noncumulative) and enhanced liquidity stress test (ELST), Basel III ratios (LCR as per revised guidance published in Jan. 2013 and NSFR based on BCBS guidance (Dec. 2010); and liquidity risk model by Wong and Hui (2009).

• TD by IMF staff: implied cash flow test (cumulative) and 30-day implied cash flow test (noncumulative), with focus on the sudden, sizeable withdrawal of funding retail deposit run.(liabilities) and the sufficiency of existing assets to withstand those shocks under stressed conditions; also maturity mismatch analysis (both local and foreign currencies).

• Also one alternative scenario (for IMF TD test and ELST), which assumes the absence of a deposit run.

• Only unencumbered liquid assets (generating cash inflows), ie, that can be sold or used as a collateral to receive funding (with the exception of cash/cash-equivalents).

• Buffer: Counterbalancing capacity and Hong Kong Monetary Authority’s liquidity facilities.

• TD by authorities: Basel III ratio (LCR as per revised guidance published in Jan. 2013 and implementation guidance issued by Swiss authorities, effective as of Jan. 1, 2016).

• TD by IMF staff: Implied cash flow test (cumulative) of different deposit run-off and asset disposal rates over a predetermined period of time (five working days); assesses the liquidity stance and counterbalancing capacity of banks at the end of each day; deposit run-off rates and asset disposal rates were based on expert judgment.

• Buffer: Counterbalancing capacity.

• BU: Assessment of the counterbalancing capacity via coverage ratios (excess liquidity coverage and LCR); structural maturity mismatch ratio via cash flow-based test using maturity buckets (funding ratio); excess liquidity coverage and funding ratios according to Section 152 DFBA and Supervisory Diamond, LCR according to CRD-IV.


• Performance of banks under stress (estimated survival period in days, number of banks which still meet their contractual obligations).

• Performance of banks under stress (how many banks fail in 5 days, or 30-day period, liquidity needs, maturity mismatch profile, and prospective Basel III ratios).

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets, amortization of outstanding assets, related party lending, and contingent claims/liabilities (undrawn/uncommitted).

• Hurdle metrics: Distribution of ratios, number of failed banks, and liquidity shortfall relative to unencumbered assets.

• Performance of banks under stress in maintaining net positive liquidity position.

• Hurdle metrics: Available liquid assets to cover liquidity needs under each scenario.

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets) in maintaining net positive liquidity position (i.e., counterbalancing capacity above potential cash outflows in stress scenario over 30-day horizon).

• Hurdle metrics: Change in net liquidity position and counterbalancing capacity; drivers of main results (liquidity position/counterbalancing capacity), for each scenario; number of banks (pass rate) below the minimum requirement, for each scenario; and differentiation of results between Italian banks (top five, large-/medium-/small-sized) and foreign banks’ subsidiaries.

• BU: Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets). Hurdle metrics: Liquidity funding gap by bank/currency, and consolidated across currencies.

• TD: Idiosyncratic and market-wide shock as described in the LCR. Hurdle metrics: Finalized LCR rules by bank/currency (and consoliated across currencies).

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets) over 45 scenarios including full/limited/restricted/closed access to money markets covering funding in local/foreign currency, euro currency, and FX swap markets.

• Hurdle metrics: Liquidity funding gap by bank/currency and percentage of assets that “fail” under each scenario (i.e., not covered by cash inflows).

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets); liquidity losses are assumed to be equal to 2.25 percent of RWAs.

• Funding and market liquidity risk (including information contagion risk) due to solvency issues.

• The “run point” serves as an indication of how bank capital, liquid assets, and the term structure of outstanding debt impact liquidity risk.

• Hurdle metrics: Expected losses from liquidity funding gap.

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets, amortization of outstanding assets, related party lending, and contingent claims/liabilities [undrawn/uncommitted]).

• Hurdle metrics: Distribution of ratios, number of failed banks, and liquidity shortfall relative to unencumbered assets; expected first cash shortage time (CST), probability of cash shortage (CSP), expected default time due to liquidity problems (LFT), and probability of default due to liquidity problems (LFP) according to the model by Wong and Hui (2009).

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets, amortization of outstanding assets).

• Hurdle metrics: Distribution of ratio (LCR); pass rate and remaining buffers (systemwide and by bank type).

• Performance of banks under stress (i.e., bank run and liabilities run-off, taking into account valuation haircuts to liquid assets).

• Funding and market liquidity risk scenarios with different shocks to market values of liquid assets and downgrades of financial institutions.

• Hurdle metrics: Distribution of ratios (systemwide and different bank size buckets); pass rate and liquidity shortfall/potential liquidity shortfall.

• B/S. • B/S. • B/S. • B/S. • B/S using IMF assumptions. • B/S using IMF assumptions.

• B/S using IMF assumptions. • B/S. • B/S using IMF templates and assumptions.• Model-based [Wong-Hui (2009) approach].

3/

• B/S. • B/S.

• Technical Note, published.





published.


• Results discussed in FSSA.


published.




published.


published.



published.


published.












L. Schumacher S. Munoz

E. Loukoianova A. Jobst S. Nowak (EUR)

J.A. Chan-Lau H. Oura E. Kopp

J.A. Chan-Lau I. Guerra

L. Valderrama I. Krznar J. Surti

A. Jobst C. Baba

C. Caceres F. Lipinsky

E. Kopp




http://www.imf.org/external/pubs/ft/scr/2013/cr13137.pd












420


Liquidity Stress Test Matrix (STeM) for FSAPs of Systemically Important Financial Systems (Illustrative)

29 G20, S29

South Africa

30 G20, S29

Korea

31 G7, G20, S29

United States

32 S29

Norway

33 G20

Argentina

34 G7, G20, S29

United Kingdom

35 G7, G20, S30

Germany

36 S29

Ireland

37 G20, S29

Russian Federation

38 S29

Finland**

39 S29

Turkey

December 2014 January 2015 July 2015 September 2015 February 2016 June 2016 June 2016 September 2016 September 2016 January 2017 February 2017

• BU by banks using templates and assumptions provided by FSAP team.

• BU by banks using supervisory (Financial Supervisory Service) templates and assumptions.

• TD by authorities (Financial Supervisory Service).

• TD by IMF staff. • TD by IMF staff. • BU by banks using templates and assumptions provided by IMF staff.

• TD by authorities (Bank of England), using IMF staff templates and assumptions.


• TD by authorities (Bundesbank) with inputs from IMF staff.

• TD by Central Bank of Ireland and IMF staff.

• TD by authorities (Central Bank of Russia), in collaboration with IMF staff.

• TD by authorities (FSA), in collaboration with IMF staff.

• BU by banks, in collaboration with authorities and IMF staff (only LCR-based reverse liquidity stress test)

• TD by authorities (FSA), in collaboration with IMF staff.

• 6 largest banks (local and foreign operations in order to adequately assess the risks associated with the rapid expansion in cross-border business).

• BU and TD: All nationwide commercial banks (7).

• TD only: All other commercial banks, on an aggregate basis.

• 31 largest bank holding companies (BHCs).

• 6 largest commercial and savings banks.

• 22 major banks • 10 large financial institutions comprising seven major banks building societies and the three largest subsidiaries of foreign investment banks; firms were selected to provide desired coverage (see below) as measured by the PRA048 liquidity returns.

• BU: 44 banks participating in the Basel Committee’s Quantitative Impact Study (QIS).

• TD: All 1,800 banks operating in Germany.

• 5 largest banks. • 681 banks. • 4 largest banking groups. • 10 largest (domestic) banks.


• BU: 76 percent of total sector assets.





• 80 percent of total sector assets. • BU: >90 percent of total sector assets.


• 61 percent of total sector assets. • 99.1 percent of total sector assets.

• 93 percent of total sector assets. • >80 percent of total sector assets.

• Banks’ own data. • BU: Banks’ own data.• TD: Supervisory data.

• Publicly available data (FR Y-9C).

• Supervisory data from FSA. • Banks’ own data. • LCR: Supervisory data from the PRA return (using interim LCR reporting on all currencies based on the EU Delegated Act [European Commission Delegated Regulation No. 2015/61], which implements the LCR in the U.K.).

• Implied cash flow test: Supervisory data based on prudential return (PRA048).

• Supervisory and banks’ own data. • Supervisory data. • Supervisory and public data. • Supervisory and banks’ own data. • Supervisory and banks’ own data.

• End-Q1 2014. • End-Q1 2013. • End-Q3 2014. • End-2014 (and end-Q3 2014).7 • End-Q3 2012. • End-2015. • BU: End-Q2 2015.• TD: End-Q2 and end-Q4 2015.

• End-Q2 2015. • End-2015. • End-2015. • End-2015 (and sensitivity analysis of data for previous years).

• Solo basis. • Consolidated banking group.

• Consolidated. • Consolidated. • Solo basis. • Solo basis. • n.a. • Solo basis. • Solo basis. • Solo basis. • Consolidated banking groups.

• BU: Basel III ratios (LCR as per revised guidance published in Jan. 2013 and NSFR based on BCBS guidance of Dec. 2010).

• Buffer: Central bank’s provision of committed liquidity facility.

• Cash flow mismatch analysis: 30-day implied cash flow test, with focus on the sudden, sizeable withdrawal of funding (deposits, wholesale funding, derivatives, and committed credit lines) and the sufficiency of selling (unencumbered) existing assets to withstand those shocks under stressed conditions; the magnitude of wholesale funding shock in line with historical experience (global financial crisis) and the deposit withdrawal rates are calibrated to be more severe than the historical experience in Korea.

• BU buffer: Counterbalancing capacity, with banks allowed to generate cash inflows via asset sales, use of excess reserves (as central bank deposits), and access to the central bank liquidity facilities.

• TD: Assesses the counterbalancing capacity of banks based on the ratio of their stock of liquid assets to the total (noncumulative) cash outflow over two different time periods (one/three consecutive quarters); deposit run-off rates were calbrated to the historical experience during the 2008/09 episode of the financial crisis, and asset disposal rates were taken from the LCR. 4,5


• Basel III ratios (LCR as per revised BCBS guidance published in Jan. 2013 and NSFR based on BCBS guidance, Oct. 2014).

• Two scenarios with varying shocks calibrated to (i) the market experience after the collapse of Lehman Brothers in 2008; and (ii) other past episodes of liquidity stresses—both scenarios were more severe than the historical experience in Norway—in order to simulate the combined effect of (a) the inability of rolling over maturing secured funding, deposit run, and withdrawal of contingent liabilities; and (b) the inability of rolling over maturing unsecured wholesale funding.


• Basel III ratios (LCR and NSFR per BCBS guidance [Dec. 2010])

• Implied cash flow test (with maturity buckets), with focus on the sudden, sizeable withdrawal of funding (i.e., bank run and dry-up of wholesale funding markets calibrated to Argentina’s experiences with banking panics during the convertibility period between 1995 and 2001, taking into account haircuts to liquid assets) and the sufficiency of existing assets to withstand those shocks under stressed conditions; run-off rates calculated following historical test.

• Macro stress tests using authorities’ macroeconomic and satellite models with FSAP team guidance.

• Reserve liquidity test for cross-validation in order to assess the capacity of banks to withstand maximum wholesale deposit withdrawals.

• Buffer: Counterbalancing capacity and central bank facilities.

• Basel III ratio (LCR) on three scenarios: (i) standard assumptions according to BCBS guidance; (ii) “U.K. retail stress” scenario: calibration of this deposit run-off scenario replicates the peak stress during the 2007 Northern Rock run, with run-off rates for retail deposits of up to 15 percent and for corporate deposits of 60 percent, and with liquidity risk from committed but undrawn liquidity facilities of 50 percent; and (iii) “U.K. wholesale stress” scenario: replicates the liquidity stress observed during the global financial crisis, characterized by a freeze of wholesale funding and liquidity risk from sizeable margin calls related to secured funding, derivatives and foreign currency funding due to market liquidity shocks, derivative assignments, and unwinds and disruptions in the FX swap market (with rollover of secured funding backed by other than Level 1 and Level 2A assets of up to 0 percent).

• Implied cash flow test (cumulative over five days and noncumulative over 30 days), with focus on the sudden, sizeable withdrawal of funding and the sufficiency of existing assets to withstand those shocks; outflow shocks are applied to a range of liabilities, including deposits, wholesale funding and intergroup funding, while haircuts to assets include investment and trading securities, derivatives and secured assets; does not consider offsetting contractual cash inflows from maturing wholesale lending and central bank support via the BoE’s discount window.

• A general maturity mismatch analysis by maturity bucket.

• A single currency analysis based on PRA’s ILG regime.


• TD: Implied cash flow test to assess resilience to multifactor scenario, with focus on the sudden, sizeable withdrawal of funding and the sufficiency of existing assets to withstand those shocks; approximates banks’ liquidity coverage ratio (LCR) under CRD IV, consistent with Regulation (EU) No. 575/2013 of the European Parliament and the European Council on prudential requirements for credit insitutions and investment firms and guidance by the BCBS (2013) on the LCR and liquidity risk monitoring pools.

• BU: Basel III ratios (LCR as per revised BCBS guidance published in Jan. 2013 and NSFR based on BCBS guidance, Oct. 2014).

• Proxy measures of the Basel III ratios (LCR as per Commission Delegated Regulation (EU) 2015/61 of Oct. 2014 and NSFR based on BCBS guidance published in Oct. 2014).

• Implied cash flow test (using maturity buckets by banks), with focus on the sudden, sizeable withdrawal of funding and the sufficiency of existing assets to withstand those shocks.

• Buffer: Counterbalancing capacity, central bank facilities.

• Implied cash flow test (sudden outflows due to liabilities run-off and sufficiency of existing assets to withstand those shocks under stressed conditions [after application of haircuts]); range of run-off rate between 10 percent (household deposits) to 50 percent (nonresident, interbank deposits); range of haircuts between 5 percent (highly liquid assets) and 65 percent (less liquid assets).


• Implied cash flow test (by maturity bucket) simulating a sudden, sizeable withdrawal of funding and the sufficiency of existing assets to withstand those shocks, after considering effect on liquid assets. • Basel III ratios (LCR as per Commission Delegated Regulation (EU) 2015/61 of Oct. 2014 and NSFR based on BCBS guidance published in Oct. 2014); for the LCR-based analysis, two alternative scenarios were examined: (i) a more severe withdrawal rate of deposits and (ii) a dry-up of unsecured wholesale funding, calibrated to meet very severe stress conditions, such as those experienced during the 2008/09 global financial crisis; also a separate LCR-based analysis of foreign currency positions and a reverse LCR-based liquidity stress test were carried out.


• Implied cash flow test (by maturity bucket) simulating a sudden, sizeable withdrawal of funding and the sufficiency of existing assets to withstand those shocks, after considering effect on liquid assets.

• Basel III ratio (LCR as per revised BCBS guidance published in Jan. 2013), for each currency and consolidated across all currencies.


• Performance of banks under stress (i.e., liabilities run-off, taking into account valuation haircuts to liquid assets, amortization of outstanding assets, related party lending, and contingent claims/liabilities [undrawn/uncommitted]).

• Hurdle metrics: Number of banks that fell within certain ratios.

• Performance of banks under stress (i.e., liabilities run-off and asset sales after valuation haircuts).

• Hurdle metrics: Liquidity shortfall; total cash in- and outlows by type of funding (indiv. institutions), deparated by local and foreign currency.


• Alternative scenario: Run-off rates calibrated based on historical experience (not LCR).

• Hurdle metrics: Distribution of ratio of the stock of liquid assets to the total cumulative cash outflow (systemwide).

• Performance of banks under stress (i.e., liabilities run-off after accounting for scheduled cash in- and outflows and the counterbalancing capacity of unencumbered liquid assets).

• Funding and market liquidity risk scenarios comprising a deposit run, dry-up of wholesale funding markets, and different shocks to market values of liquid assets.

• Hurdle metrics: Distribution of ratio (LCR/NSFR); liquidity gaps.


• Alternative scenario: Run-off rates calibrated based on historical experience (not LCR).

• Hurdle metrics: Liquidity gap, i.e., distribution of ratio of the stock of liquid assets to the total cumulative cash outflow (systemwide) and survival period in days by bank.

• LCR: Assessment of the short-term resilience of banks to sudden, sizeable withdrawals of funding (liabilities) consistent with PRA’s transitional arrangement for the LCR ratio, which is more front-loaded than that prescribed by the CRR (Art. 460).

• Implied cash flow test: post-shock net liquidity position and counterbalancing capacity above net cash outflows under stress scenario.

• Changes in average liquidity position and counterbalancing capacity for each scenario.

• Liquidity ratios, disaggregated by type and size of bank for give set of idiosyncratic and marketwide shock as described in the LCR [BU].

• Hurdle metrics: liquidity funding gap by bank [joint TD].

• Liquidity gap by bank (and aggregated).

• Survival period in days by bank; number of banks that can still meet their obligations.

• Amount of liquidity deficit and number of banks with liquidity deficit.

• Liquidity gap by bank (and aggregated).

• Survival period in days by bank; number of banks that can still meet their obligations.

• For LCR: bank-specific and sector-wide stressed LCR ratios, by currency and overall, by type of banks (small/large).

• Liquidity gap by currency and aggregated, by time band and type of bank (large/small).

• B/S. • B/S. • B/S. • B/S. • n.a. • B/S using IMF templates and assumptions.

• B/S. • B/S. • B/S. • B/S using authorities’ templates and assumptions.

• B/S.








published.


published.



published.


published.


published.


published.

• No Technical Note.• Results discussed in FSSA,

published.










https://www.imf.org/~/media/Files/Publications/CR/2017/cr1706.ashx

http://www.imf.org/~/media/Files/Publications/CR/2017/cr1735.ashx

L. Lin S. Iorgova P. Jeasakul H. Wong

I. Krznar B. Huston F. Lipinsky

S. Iorgova M. Catalan L. Valderrama E. Kopp H. Kang C. Pouvelle

A. Alter N. Belhocine R. Cervantes F. Lipinsky

C. Pouvelle N. Valkx











https://www.imf.org/~/media/Files/Publications/CR/2017/cr1706.ashx

http://www.imf.org/~/media/Files/Publications/CR/2017/cr1735.ashx

Andreas A


hristian Schmieder

421

Source: Compiled by authors with contributions from respective FSAP stress testers.Note: The table presented here is for illustration only—the full-sized matrix is available as an MS Excel® file (entitled Stress Testing Matrix [STeM] − Liquidity Stress Testing Approaches in IMF FSAPs) on the IMF eLibrary at www.elibrary.imf.org/page/stress-test2-toolkit. BCB = Banco Central do Brasil; BCBS = Basel Committee on Banking Supervision; BoE = Bank of England; BoJ = Bank of Japan; BU = bottom-up; CRR = Capital Requirements Regulation; ECB = European Central Bank; ELST = enhanced liquidity stress test; FSSA = Financial System Stability Assessment; FX = foreign exchange; HKMA = Hong Kong Monetary Author-ity; LCR = liquidity coverage ratio; NBB = National Bank of Belgium; NPL = nonperforming loan; NSFR = net stable funding ratio; MFRAF = Macro-Financial Risk Assessment Framework; PRA = Prudential Regulatory Authority; TD = top-down.* Staff from the Monetary and Capital Markets Department (MCM) of the IMF unless specified otherwise.** Four additional countries (Denmark, Finland, Norway and Poland) were added to the original S-25 list following the 2013 decision of the IMF's Executive Board (IMF 2014a). At the time of the FSAP, Finland was not a S-29 country (subject to the mandatory five-year FSAP cycle). Since the 2016 FSAP for Finland has not been completed by end-September 2016, the description of the last FSAP (2010) is shown here.1 Basel Committee on Banking Supervision (2013a).2 Basel Committee on Banking Supervision (2010b).3 The CDP is a specialized lending entity, which is majority-owned by the Italian government and funds itself mostly by postal and customer deposits. It is required to deposit the liquidity provided by postal sav-ings with the Italian Treasury, which makes up nearly half of total assets.4 Each outflow item is a sum of the corresponding outflow item across 31 bank holding companies. The run-off rate is then defined as the percentage difference between the value of a given outflow items at its peak and its value next quarter or three quarters from the peak.5 High-quality liquid assets in the US final rule on the LCR does not include securities issued or guaranteed by public sector enterprises (for example, state, local authority, or other governmental subdivision below the sovereign level), such as municipal securities or residential mortgage-backed securities (RMBS). Moreover, claims issued or guaranteed by a US government–sponsored enterprise (GSE) are not included in Level 1 liquid assets and corporate debt securities are not included in Level 2A assets.6 The narrow banking system includes city, trust, regional (Tier 1 and 2), foreign, bridge, and internet banks as well as shinkin banks and credit cooperatives. The broader system also includes J-Post bank and Norinchukin bank.7 The end-2014 data were based on the high-quality liquid asset (HQLA) definition (and related haircuts) of the European Commission's Delegated Act on the LCR (Oct. 2014), whereas the LCR data as of end-Sept. 2014 were based on the original Basel III definition. The use of two alternative specifications was motivated by the need to evaluate the impact of the broader definition of HQLA under EU regulations (by including highly rated covered bonds) on the LCR.


www.elibrary.imf.org/page/stress-test2-toolkit. BCB = Banco Central do Brasil; BCBS = Basel Committee on Banking Supervision; BoE = Bank of England; BoJ = Bank of Japan; BU = bottom-up; CRR =

Appendix 16.2.Funding and Market Liquidity

For funding liquidity risk, the assessment reflects the realization (and potential change) of expected and contingent cash in- and outflows during times of stress, which includes assumptions about:

• Runoff rates for secured/unsecured wholesale and retail funding.• Amortization/renewal rates for secured/unsecured wholesale and retail lending (at contractual maturities).• Draw- down rates for interbank credit and liquidity facilities.• The convertibility of foreign currency- denominated net cash flows and the scope of unsecured support in convertible

currencies from related and third parties in the form of committed/uncommitted lines.• The treatment of expected and contingent liabilities from related and third parties.• The capacity to access unsecured financing and complete securitization during times of stress.

The degree of market liquidity risk (that is, valuation haircuts) affecting expected cash inflows from asset sales and the col-lateralization of secured funding are influenced by:

• The asset concentrations and banks’ asset encumbrance.• The potential impact of downgrades of marketable assets.• The composition of the bank’s liquidity buffer comprising marketable, or otherwise realizable, assets.• The magnitude of foreign currency funding needs— on aggregate and for each currency (if there is no full convertibility

between currencies over the stress testing time horizon).• The relevance of derivatives trading for the management of liquidity risk, including asset and foreign currency swaps

(with the attendant potential for collateral and margin calls).• The extent to which assets might be encumbered and are subject to haircuts when used as collateral for central bank and

securities financing transactions during times of stress, such as repos and securities lending.• The availability of funding via potentially reusable securities received as collateral (“rehypothecation”).


Appendix 16.3. Regulatory Liquidity Risk Measures under Basel III: Liquidity Coverage Ratio (LCR) and Net Stable Funding

Ratio (NSFR)

In the wake of the global financial crisis, the Basel Committee on Banking Supervision (BCBS) added liquidity risk to the regulatory perimeter of the Basel III framework (BCBS 2009, 2010a, 2010b, 2017d). Internationally active banks must meet two quantitative liquidity metrics (and related monitoring tools) for two different time horizons (one month and one year, re-spectively) (Appendix Table 16.3.1) and comply with qualitative guidance liquidity riskmanagement practices. As such, banks are expected to maintain a stable funding structure to withstand liquidity shocks by holding a sufficient stock of assets that should be available to meet their funding needs in times of stress and by limiting maturity transformation (BCBS 2011, 2012b).17 Recent work by the BCBS (2016b) sought to shed light on effects of the liquidity reforms under Basel III and their interaction with capital standards.

Liquidity Coverage Ratio (LCR)

The LCR is intended to promote short- term resilience to potential liquidity shocks by requiring banks to hold a sufficient stock of unencumbered, high- quality liquid assets (HQLAs) to withstand a liabilities runoff over a stressed 30-day scenario specified by supervisors. The potential funding shortfall is defined as cash outflows less cash inflows (subject to a cap of 75 percent of total ex-pected cash outflows). A liquidity coverage ratio value of less than 100 percent indicates a liquidity shortfall. More specifically,

“… the LCR numerator consists of a stock of unencumbered, high- quality liquid assets that must be available to cover any net [cash] outflow, while the denominator is comprised of cash outflows less cash inflows (subject to a cap at 75 [percent] of total outflows) that are expected to occur in a severe stress scenario” (BCBS 2012b, 2013a).

In January 2013, BCBS finalized the specification of the LCR by reaching an agreement on a composition of HQLAs and parameters for net cash outflows resulting from deposits and contingent liabilities, as well as a transition period for introduc-tion of the LCR (BCBS 2013a). The changes to the definition of the LCR include an expansion in the range of assets eligible as HQLAs and some refinements to the assumed inflow and outflow rates to better reflect actual experience in times of stress. More specifically, the modifications comprise the following:18

• Extending the Level 2B category of the HQLA to include (1) residential mortgage- backed securities (rated “AA” and higher) with a haircut of 25 percent as well as lower- rated corporate bonds (between “A+” and “ BBB-”) and common equity (each subject to a 50 percent haircut); and increasing the cap of Level 2B assets from 10 to 15 percent.

• Applying a lower runoff rate of 3 percent to stable deposits where prefunded and explicitly government- guaranteed deposit insurance schemes exist and where access to deposits is available the next day after deposit insurance is triggered.

• Lowering the drawdown rates from 100 to 30 percent for undrawn but committed liquidity facilities to nonfinancial firms, sovereigns and central banks, public sector enterprises, and multilateral development banks19 from 100 to 40 per-cent for undrawn but committed credit/liquidity facilities to banks subject to prudential supervision, and from 75 to 40 percent for deposits from nonfinancials, sovereigns, and public- sector enterprises.

17 See Bucalossi and others 2016 for a detailed analysis of the potential impact of standardized liquidity risk measures on banks’ liquidity management in the European context.

18 See also BCBS 2014.19 For committed credit facilities the drawdown rate declines to 10 percent. The assumed drawdown rate for both credit (liquidity) facilities extended to

other nonbank financial institutions including securities firms, insurance companies, fiduciaries, and beneficiaries is 40 (100) percent (BCBS 2013a).



• Increasing liquidity needs related to derivatives.• Applying a zero percent haircut/discount factor for operations with central banks for all types of assets (in addition to

secured funding backed by Level 1 assets with any counterparty).• Providing for national treatment of trade finance obligations.

Net Stable Funding Ratio (NSFR)

The NSFR restricts liquidity mismatches from excessive maturity transformation to encourage longer term borrowing by limit-ing the stock of unstable funding (BCBS 2014, 2016). It was implemented on January 1, 2018, following an observation period that included a review clause to address any unintended consequences. Based on the current definition, banks are required to establish a stable funding profile over the short term, that is, the use of stable ( long- term and/or stress- resilient) sources to con-tinuously fund cash- flow obligations that arise from lending and investment activities inside a one- year time horizon.

The NSFR reflects the proportion of longer term (and less liquid) assets that are funded by stable sources of funding, includ-ing customer deposits, wholesale funding with maturities of more than one year, and equity (but excludes short- term liabili-ties). These sources and uses of funds are not equally weighted but enter as risk- adjusted components into the calculation of the NSFR. A value of this ratio of less than 100 percent indicates a shortfall in stable funding based on the difference between balance sheet positions after the application of available stable funding factors and the application of required stable funding factors for banks where the former is less than the latter (BCBS 2010c, 2014b).20

20 Compliance with the NSFR, which emphasizes the availability of long- term sources of funding, could conflict with plans to make senior bondholders absorb bank losses under so- called “ bail- in” clauses (Pengelly 2012). Banks might find it difficult to lengthen the maturity of their balance sheet by issu-ing additional unsecured debt if mandatory bail- in clauses were attached to them, which would also result in higher funding costs compensating for in-vestors for accepting bail- in risk.


Andreas A


hristian Schmieder

425


Overview of the Basel II and III Minimum Capital Requirements and Liquidity StandardsYear 2011 2012 2013 2014 2015 2016 2017 2018 2019

Leverage Ratio Supervisory Monitoring

Parallel Run January 2012–January 2017; Disclosure Started in January 2015

Migration to Pillar I

Minimum Common Equity Capital Ratio 3.50 4.00 4.50 4.50 4.50 4.50 4.50Capital Conservation Buffer 0.625 1.25 1.875 2.50Minimum Common Equity + Capital Conservation Buffer 3.50 4.00 4.50 5.125 5.75 6.375 7.00Phase-in of Deductions from Core Equity Tier 1 20 40 60 80 100 100Minimum Tier 1 Capital 4.50 5.50 6.00 6.00 6.00 6.00 6.00Minimum Total Capital 8.00 8.00 8.00 8.00 8.00 8.00 8.00Minimum Total Capital + Capital Conservation Buffer 8.00 8.00 8.00 8.625 9.25 9.875 10.50Phase-out of instruments that no longer qualify as non-Core

Tier 1 or 2 capitalPhased out over a 10-year horizon beginning 2013

Liquidity Coverage Ratio Start of obs.

period


60 70 80 90 100Net Stable Funding Ratio Start of

obs.period


Source: Basel Committee for Banking Supervision (BCBS), http://www.bis.org/bcbs/basel3.htm (BIS 2017). Note: See BCBS (2010b and 2010c). The introduction of the Liquidity Coverage Ratio (LCR) was graduated (BCBS 2013a). Specifically, the LCR was introduced on January 1, 2015, but the minimum requirement began at 60 percent, rising in equal annual steps of 10 percentage points to reach 100 percent on January 1, 2019. Obs = observation.


http://www.bis.org/bcbs/basel3.htm (BIS 2017

Appendix 16.4.The Interaction and Integration of Solvency and Liquidity Risks

More comprehensive macroprudential stress tests incorporate feedback effects between solvency conditions and liquidity risk in banking sectors. While several papers have taken a systematic approach for analyzing their interaction, the practical implemen-tation of this concept remains at an early stage (BCBS 2013c, 2015). For example:

• Van den End (2008) develop a stress testing model that endogenizes market and funding liquidity risks by including feedback effects, which capture both behavioral and reputational effects.

• Aikman and others (2009) integrate funding liquidity risk and solvency risk in the Bank of England’s Risk Assessment Model for Systemic Institutions. The framework simulates banks’ liquidity positions conditional on their capitalization under stress, and other relevant dimensions such as a decrease in confidence among market participants under stress.

• Wong and Hui (2009) explicitly capture the link between default risk and deposit outflows. Their framework simulates the impact of mark- to- market losses on banks’ solvency positions leading to deposit outflows; asset fire sales by banks are evaporating and contingent liquidity risk sharply increases.

• Barnhill and Schumacher (2011) develop a more general empirical model, incorporating the previous two approaches that attempt to be more comprehensive in terms of the source of the solvency shocks and compute the long- term impact of funding shocks.

• Schmieder and others (2012) construct an Excel- based tool that allows liquidity tests informed by banks’ solvency conditions and simulates the increase in funding costs resulting from deteriorating solvency.

• Jobst (2014) combines option pricing with market data and balance sheet information in the Systemic Risk- adjusted Liquidity model to generate a probabilistic measure of the frequency and severity of multiple entities experiencing a joint liquidity event. The model links a bank’s maturity mismatch between assets and liabilities affecting the stability of its funding with the characteristics of other banks, subject to individual changes in risk profiles and common changes in market conditions.

• Anand, Bédard- Pagé, and Traclet (2014) include a top- down liquidity stress test in the Bank of Canada’s Macro- Financial Risk Assessment Framework, which takes into account additional sources of pressure of banks’ solvency due to outright rationing of funding— in addition to increases in its cost— and secondary effects from potential spillovers with counterparty risk as weak banks may be unable to honor, in part or entirely, their interbank exposures.

• Hesse, Salman, and Schmieder (2014) integrate macro- financial linkages, namely spillovers from the European periph-ery, to banks’ solvency and liquidity resilience in a stress testing framework.


Appendix 16.5.Liquidity Stress Testing Using Implied

Cash Flows

Over the years, the IMF staff has developed several liquidity risk stress testing tools for the system- wide assessment of the im-pact of negative shocks to banks’ funding conditions. This appendix presents one of these tools (Jobst 2017), which was applied in the financial stability assessment modules of the Financial Sector Assessment Programs (FSAPs) for Hong Kong SAR (IMF 2014d) and the United Kingdom (IMF 2016b).21 The tool provides instructions regarding data requirements and assumptions and contains a complete calculation methodology consistent with the specific liquidity stress testing requirements of FSAPs.

The liquidity stress test captures the risk of a bank failing to generate sufficient funding to satisfy its short- term payment obligations over a predefined stress horizon. It follows a top- down implied- cash- flow (ICF) approach of modeling the impact of the sudden, sizeable withdrawals of funding (that is, liabilities runoff) and unscheduled cash outflows after taking into account the repayment of outstanding claims and availability of existing liquidity buffers (“counterbalancing capacity”). The funding shock is calibrated to assumptions about the expected (that is, scheduled) and potential cash inflows and outflows related to existing claims and obligations (“funding liquidity risk”) and the application of haircuts to available assets (“market liquidity risk”) over risk horizons of five days (cumulative) and 30 days (noncumulative). The ability to survive funding constraints is also influenced by the degree to which saleable assets are encumbered and the rollover risk stemming from maturity mis-matches of assets and liabilities, which are assessed for both local and foreign currencies.

More specifically, several channels affecting the severity of cash- flow calculations are considered (Appendix Table 16.5.1). They comprise: (1) the decline in asset values under stress and the extent to which they can be either used as collateral for secured wholesale funding or sold at stressed market values (“market liquidity risk”), (2) callback/renewal rates of scheduled and unsched-uled cash flows from maturing assets and liabilities (“funding liquidity risk”), and (3) the utilization rate of contingent claims and liabilities/funding swap arrangements.22 More specifically:

• Liquid assets available for sale or collateralized funding under the assumption of varying degrees of asset- specific valuation haircuts and encumbrance levels comprise: (1) cash and cash balances with central banks, (2) securities and bank loans eligible for refinancing operations at the domestic and major central banks, (3) securities and bank loans that can be mobi-lized in repo transactions (or another type of lending against financial collateral), and (4) marketable securities in general.

• Cash inflows are determined by the expected repayment amount of outstanding credit with/without liquid financial assets as collateral, comprising: (1) expected inflows of cash and decline of liquid assets related to maturing transactions with/without liquid securities and bank loans (for example, repo and securities lending transactions), (2) expected and potential net cash flows related to derivatives (excluding credit derivatives), and (3) potential inflows from committed/ uncommitted credit lines to related and third parties.

• Cash outflows are defined by the runoff of maturing and nonmaturity funding with/without liquid financial assets as collateral, comprising: (1) expected inflows of cash and increase of liquid assets related to transactions with/without liquid securities and bank loans (for example, reverse repo and securities borrowing transactions), (2) maturing repay-ments to related parties, and (3) committed/uncommitted contingent claims to related and third parties.

The liquidity stress test is evaluated numerically as the ratio between potentially available liquidity and potentially required liquidity, which should be at least 100 percent or greater. A value lower than 100 percent would imply a liquidity shortage if the assumed stress scenario materialized. The test also includes several additional assumptions:

• Only unencumbered liquid assets (generating cash inflows), that is, assets used as collateral to receive funding (except for cash/ cash- equivalents), are included in the test (“liquidity scope”). Funding via potentially reusable securities received

21 The tool is available as an MS Excel• file (entitled “IMF FSAP Liquidity Stress Testing Tool”) on the IMF eLibrary at https://www.elibrary.imf.org/page /stress-test2-toolkit.

22 The workbook requires firm- level data on liquid assets, inflows and outflows from specified assets and liabilities, and net flows from derivatives, which are sepa-rated into two “maturity buckets” of either: (1) one week/open maturity; or (2) longer than one week but up to one month, corresponding to the respective implied- cash- flow tests. The five- day test includes only data provided for the first maturity bucket, which are subject to the cumulative impact of specific call-back and runoff rate assumptions of assets and liabilities. The assumptions on valuation haircuts (for liquid assets), callback rates (for cash inflows from the rolloff of outstanding claims and potential funding from contingent liabilities), and runoff rates (for cash outflows from the withdrawal/termination of funding and potential payments from contingent claims) are organized in separate worksheets and can be amended according to country- specific circumstances.





as collateral (“rehypothecation”) and cash inflows from new or renewed (secured/unsecured) wholesale lending (at con-tractual maturities) but full renewal of secured retail lending (for example, secured lending with illiquid collateral such as residential mortgages) are not considered.

• There is limited potential unsecured support in convertible currencies from related and third parties (for example, in the form of committed line) but full convertibility between currencies (within one week).

In the recent FSAP for the United Kingdom (IMF 2016b), for example, the liquidity stress testing tool was applied to 10 institutions, consisting of seven major commercial banks and building societies, and the three largest subsidiaries of foreign investment banks covering 80 percent total banking assets. Results for the five- day and 30-day ICF tests suggest that the pro-tracted noncumulative cash flows over a longer time horizon have a greater impact on the banks’ liquid buffers than cumulative stresses over a shorter period (Appendix Table 16.5.2 and Appendix Figure 16.5.1).


Andreas A


hristian Schmieder

431


Liquidity Stress Test Tool—Summary of Assumptions

Test Definition

Basic Assumptions

Other AssumptionsAssets (cash inflows) Liabilities (cash outflows)

Five-day implied-cash-flow test

Cumulative inflow and outflow over five consecu-tive days

Liquid financial assets: (1) cash and cash balances with central banks [haircut: 0 percent], (2) securities and bank loans eligible at major central banks [0–15], (3) securities and bank loans that can be mobilized in repo transactions (or another type of lending against financial collateral) [5–30], and (4) marketable securities [10–35].

Cumulative cash outflows: (1) maturing and nonmaturity funding without liquid financial assets as collateral [runoff rate: 5 percent per day] (that is, all deposits and funding from financial and nonfinancial corporates as well as private households and SME clients) with the exception of sovereign and other public sector and central bank clients [0], (2) expected outflows of cash and liquid assets related to transactions with liquid securities and bank loans (for example, reverse repo and securities borrowing transactions) [20], (3) maturing outflows to related parties [20], and (4) committed/uncommitted contingent claims to related and third parties [drawdown rate: 3/5 percent per day].

A ratio lower than 100 percent implies a liquidity shortage if the stress scenario would materialize at the reporting date (that is, potentially required liquidity > potentially available liquidity); only unencumbered liquid assets (generating cash inflows), that is, assets used as collateral to receive funding (with the exception of cash/cash-equivalents) are included in the test (“liquidity scope”); new unsecured financing and securitiza-tion impossible within the time horizon; no offsetting cash inflows from new or renewed (secured/unsecured) wholesale lending (at contractual maturities) but full renewal of secured retail lending (for example, secured lending with illiquid collateral [residential mortgages]); central bank eligible collateral can be monetized at appropriate haircuts; repo markets are open at appropriate haircuts; firesale of assets possible at appropriate haircuts; no consideration of funding via potentially reusable securities received as collateral (“rehypothecation”); limited potential unsecured support in convertible currencies from related and third parties (for example, in the form of committed lines); no renewal of term retail and wholesale deposits; and full convertibility between currencies (within one week).

Cumulative cash inflows: (1) expected cash inflows related to credit extension without liquid financial assets as collateral [callback rate: 20 percent per day], (2) expected inflows of cash and liquid assets related to maturing transactions with liquid securities and bank loans (for example, repo and securities lending transactions) [20], and (3) potential inflows from committed/uncommitted credit lines to related and third parties [drawdown rate: 3/5 percent per day].

Cumulative net cash flows: expected and potential net cash flows related to derivatives (excluding credit derivatives)—net contractual cash flows [20]1

(continued)





432


Liquidity Stress Test Tool—Summary of Assumptions

Test Definition

Basic Assumptions

Other AssumptionsAssets (cash inflows) Liabilities (cash outflows)30-day

implied-cash- flow test

Noncumulative Liquid financial assets: (1) cash and cash balances with central banks [0], (2) securities and bank loans eligible at major central banks [0–20], (3) securities and bank loans that can be mobilized in repo transactions (or another type of lending against financial collateral) [10–60], and (4) marketable securities [20–70].

Noncumulative cash inflows: (1) expected cash inflows related to credit extension without liquid financial assets as collateral [callback rate: 100 percent], (2) expected inflows of cash and liquid assets related to maturing transactions with liquid securities and bank loans (for example, repo and securities lending transactions) [100], (3) expected and potential net cash flows related to deriva-tives (excluding credit derivatives)—net contractual cash flows [100], and (4) potential inflows from committed/uncommitted credit lines to related and third parties [drawdown rate: 23/12 percent].

Noncumulative cash outflows: (1) maturing and nonmaturity funding without liquid financial assets as collateral [runoff rate: 10–75 percent] (that is, all deposits and funding from financial and nonfinancial corporates as well as private households and SME clients) with the exception of sovereign and other public sector and central bank clients [0], (2) expected outflows of cash and liquid assets related to transactions with liquid securities and bank loans (for example, reverse repo and securities borrowing transactions) [100], (3) maturing outflows to related parties [100], and (4) committed/uncommitted contingent claims to related and third parties [drawdown rate: 12/23 percent].

Noncumulative net cash flows: expected and potential net cash flows related to derivatives (excluding credits derivatives)—net contractual cash flows [100]1

Source: Jobst, Ong, and Schmieder 2017. Note: SME = small- and medium-sized enterprise.1Note that many derivatives positions might be non-deliverable (typically, foreign exchange and interest rate swaps and forwards), and their valuation tends to be highly variable based on prevailing market conditions and expectations. For these positions, the valuation using the firm’s chosen accounting treatment should be considered, and potential net cash flows (variation margin/cash settlement cost) would need to be checked for consistency with the calibration of market risk under the Basel framework.


Andreas A


hristian Schmieder

433


Liquidity Stress Test Results—Implied-Cash-Flow Tests (In billions of pounds sterling)Test 1a: Implied-Cash-Flow Test (Five Days)

Cumulative Loss of Unsecured Funding

(up to one week) (percent)

Cumulative Loss of Secured Funding (up to one week)

(percent)

Minimum Number of Days

of Survival

Banks Illiquid

(number)

Banks Illiquid (percent of

banking sector assets)

Net Cash Shortfall Relative to Total

Liquid Assets (percent)

Net Cash Shortfall Relative to Total Assets (percent)

Day 1 5.2 5.4 1 0 0 0 0Day 2 10.6 10.2 2 0 0 0 0Day 3 16.4 14.5 3 0 0 0 0Day 4 22.4 18.5 4 0 0 0 0Day 5 31.5 24.4 5 0 0 0 0

Test 1b: Implied-Cash-Flow Test (30 Days)

Cumulative Loss of Unsecured Funding

(percent)

Cumulative Loss of Secured Funding

(percent)

Survival Banks Illiquid

(number)

Banks Illiquid (percent of

banking sector assets)

Net Cash Shortfall Relative to Total

Liquid Assets (percent)

Net Cash Shortfall Relative to Total Assets (percent)

30 Days 27.5 100.0 No 0 0.0 0.0 0.0

Source: Bank of England staff estimates.



0

450

100

200

300

400

350

50

150

250

Five-day test(cumulative)

Alternative scenario:no retail deposit run

30-day test(noncumulative)

Alternative scenario:no retail deposit run

Sources: Bank of England and IMF staff estimates.Note: The sample of banks included in the IMF top-down implied-cash-flow stress test includes the seven largest UK banks and three large subsidiaries of foreign banks representing 80 percent of the banking sector based on a Prudential Regulation Authority liquidity reporting. Boxplots include the mean (yellow dot), the 25th and 75th percentiles (grey box, with the change of shade indicating the median), and the 10th and 90th percentiles (whiskers). The green line indicates the lowest acceptable ratio value of 100 percent (threshold).

Appendix Figure 16.5.1 Implied- Cash- Flow Tests— Distribution(In percent, solo basis)


Appendix 16.6.Liquidity Stress Testing: Reporting

Template





436

27

t0Day 1Day 2Day 3Day 4Day 5

[Country]: Aggregate Outcome of Liquidity Analysis

Test 1: Implied-Cash-Flow Analysis

Test 1a: Implied-Cash-Flow Test (5 Days)

Total Number of Banks

Cumulativeloss of allunsecured

funding(In percent)

Cumulativeloss of allsecuredfunding

(In percent)

Minimumnumberof days

ofsurvival

No. ofbanksilliquid

Percent ofbanksilliquid

(In percent)

Net cashshortfall

relative tototal liquid

assets(In percent)

Net cashshortfallrelativeto totalassets

(In percent)

Weightedavg. capitaladequacy

ratio offailingbanks

(In percent)

Weightedavg. Tier 1

capital ratio offailingbanks

(In percent)

Weightedavg. CET1

capitalratio offailingbanks

(In percent)0.0 0.0 0 0 0.0 0.0 0.01.9 16.1 1 0 0.0 0.0 0.0 0.0 0.0 0.03.7 29.0 2 16 59.3 –4.9 –0.6 6.7 7.8 9.05.4 39.4 3 27 100.0 –36.2 –4.3 12.2 14.2 16.27.0 47.6 4 27 100.0 –65.0 –7.7 12.2 14.2 16.28.5 54.2 5 27 100.0 –89.8 –10.6 12.2 14.2 16.2

Total Sample Group 1 Group 2Number of Banks failing the test 27 19 8Liquidity Shortfall –3,210,916 –2,367,750 –843,166Liquidity Shortfall (In % of total assets) –10.6 –11.6 –8.6Liquidity Shortfall (In % of liquid assets) –89.8 –98.9 –71.4

Appendix Figure 16.6.1 Example of Output Template Provided to Authorities


Andreas A


hristian Schmieder

437

Survival

Top 10% (i.e.,largest banks) 10%–25% 25%–50% 50%–75% 75%–90% 90%–100% max min

Shortfall (abs) –340,617 –421,583 –674,975 –852,080 –518,143 –403,517 –138,346 –100,426Percent of Liquid Assets 74.9 72.2 81.5 95.4 153.0 84.9

Number 3 4 6 7 4 3 27

–5,447,709 –3,928,110 –1,519,600–18.1 –19.3 –15.5

–152.4 –164.2 –128.7

Liquidity shortfall by percentile (according to assets) Individual

Top 10% (i.e.,largest banks) 10%–25% 25%–50% 50%–75% 75%–90% 90%–100% max min

Shortfall (abs) –591,310 –1,010,493 –1,765,809 –2,520,305 –2,512,344 –646,110 –218,850 –185,610

Percent of Liquid Assets 130.0 173.0 213.3 282.2 742.0 136.0Number 3 4 6 7 4 3 27

Liquidity shortfall by percentile (according to assets) Individual

Test 1b: Implied-Cash-Flow Test (30 Days)

Cumulativeloss of allunsecured

funding(In percent)

Cumulativeloss of allsecuredfunding

(In percent)

No. ofbanksilliquid

Percent ofbanksilliquid

(In percent)

Net cashshortfall

relative tototal liquid

assets(In percent)

Net cashshortfallrelativeto totalassets

(In percent)

Weightedavg. capitaladequacy

ratio offailingbanks

(In percent)

Weightedavg. Tier 1

capital ratio offailingbanks

(In percent)

Weightedavg. CET1

capitalratio offailingbanks

(In percent)15.2 100.0 No 27 100.0 –152.4 –18.1

Yes 0 0.0 0.0 0.0 0.0 0.0 0.012.2 14.2 16.2

Total Sample Group 1 Group 2Number of Banks failing the test 27 19 8Liquidity ShortfallLiquidity Shortfall (In % of total assets)Liquidity Shortfall (In % of liquid assets)






438

Bucket

No. ofbanks with

shortfall

Shortfall(In percent

of totalassets inbucket)

No. ofbanks with

shortfall

Shortfall(In percent

of totalassets inbucket)

Number offailing banks

(shortfall inlowest twomaturitybuckets)


(shortfallin any

maturitybucket)


(shortfall inlowest twomaturitybuckets)


(shortfallin any

maturitybucket)

All banks less than one week and no maturity 27 20.0 27 20.01 to 4 weeks 27 20.0 27 20.01 to 3 months 27 25.0 27 25.03 to 6 months 27 33.3 27 33.36 months to 1 year 27 50.0 27 50.0more than 1 year 27 60.0 27 60.0

Group 1 less than one week and no maturity 19 20.0 19 20.01 to 4 weeks 19 20.0 19 20.01 to 3 months 19 25.0 19 25.03 to 6 months 19 33.3 19 33.36 months to 1 year 19 50.0 19 50.0more than 1 year 19 60.0 19 60.0

Group 2 less than one week and no maturity 8 20.0 8 20.01 to 4 weeks 8 20.0 8 20.01 to 3 months 8 25.0 8 25.03 to 6 months 8 33.3 8 33.36 months to 1 year 8 50.0 8 50.0more than 1 year 8 60.0 8 60.0

Test 2b. FX onlyTest 2a. Total B/S Test 2a. Total B/S Test 2b. FX only

Test 2: Maturity Mismatch Analysis

27 27 27 27

Source: Jobst 2017.Note: This summary table was taken from the liquidity stress testing tool presented in Appendix 16.5. abs = absolute; avg = average; B/S = balance sheet; CET1 = Common Core Equity Tier 1; FX = foreign exchange; max = maximum; min = minimum; No = number.



Appendix 16.7. Cash- Flow- Based Liquidity Stress Tests

Fully fledged cash- flow- based liquidity stress tests have been implemented by several central banks and the internal liquidity risk assessment by the European Banking Authority in 2011 (and ever since then). They offer distinct advantages compared to “stock approaches” underpinning standard liquidity ratios (such as the liquidity coverage ratio [LCR]). They:

• Are forward- looking by including banks’ contractual cash outflows and inflows as well as banks’ expected counterbal-ancing capacity and should benefit from enhanced data availability and disclosure especially with regard to, for in-stance, asset encumbrance and securities funding such as repos or off- balance sheet funding.

• Enable detailed liquidity analysis and hence are better suited for capturing a bank’s funding resilience and its liquidity risk bearing capacity compared to the rather limited stock approach (IMF 2013b).

• Better capture banks’ cumulative cash flows. Standard measures follow a noncumulative approach by focusing on a specific stress test window without accounting for other detailed maturity buckets (for example, 30 days in the LCR case).

The Basel III regime23 has moved toward cash- flow- based liquidity monitoring and reporting through the LCR require-ment. Cash- flow- based liquidity stress tests have several advantages compared to other approaches. These include:

• Providing a more detailed analysis of liquidity positions similar to those carried out by banks (often daily) for their in-ternal riskmanagement purposes. The cash- flow approach incorporates securities flows and ensures consistency be-tween cash flows and securities flows.24

• Allowing for more granular maturity buckets (and may also be adapted to accommodate different currencies).• Integrating granular information on banks’ asset encumbrance levels from secured wholesale funding.• Accommodating off- balance sheet activities, such as FX swaps or credit liquidity lines, and banks’ behavioral cash out-

flow and inflows, which might be more difficult in a standard stock approach.Weaknesses of the cash- flow approach include the high data intensity as well as initial setup costs. A key prerequisite to

carry out cash- flow- based liquidity tests is access to a wide range of data on contractual cash flows for different maturity buck-ets and possibly behavioral data based on banks’ financial/funding plans. Additionally, while banks typically use a cash- flow- based approach for internal liquidity monitoring and liquidity stress testing, regulatory liquidity ratios are often based on stock accounting data with often less data granularity than the cash- flow- based templates.25

23 For instance, the Österreichische Nationalbank (OeNB) uses a cash- flow- based liquidity stress approach. Given the implementation of Basel III via the CRR/ CRD- IV framework in the European Union, uniform cash- flow templates for liquidity reporting/stress testing are likely to become a standard in other jurisdictions as well.

24 This is especially important given the fundamental role unsecured and secured wholesale funding play for many large banks.25 For EU banks, the phase- in of cash- flow- based maturity mismatch templates by the European Banking Authority provides regulators and banks with a

standardized reporting format.



Governors and Heads of Supervision.” Bank for International Settlements, Basel, Switzerland, January 8. http://www.bis.org /press/p120108.htm.

———. 2013a. “Basel III: The Liquidity Coverage Ratio and Li-quidity Risk Monitoring Tools.” BCBS Publication 238, Bank for International Settlements, Basel, Switzerland, January. https://www.bis.org/publ/bcbs238.htm.

———. 2013b. “Monitoring Tools for Intraday Liquidity Manage-ment.” BCBS Publication 248, Bank for International Settle-ments, Basel, Switzerland, April. https://www.bis.org/publ /bcbs248.htm.

———. 2013c. “Liquidity Stress Testing: A Survey of Theory, Em-pirics and Current Industry and Supervisory Practices.” BCBS Working Paper 24, Bank for International Settlements, Basel, Switzerland, October. https://www.bis.org/publ/bcbs_wp24 .htm.

———. 2013d. “Literature Review of Factors Relating to Liquidity Stress—Extended Version.” BCBS Working Paper 25, Bank for International Settlements, Basel, Switzerland, October. https://www.bis.org/publ/bcbs_wp25.htm.

———. 2014. “Basel III: The Net Stable Funding Ratio.” BCBS Publication 295, Bank for International Settlements, Basel, Swit-zerland, October. https://www.bis.org/bcbs/publ/d295.htm.

———. 2015. “Making Supervisory Stress Tests More Macropru-dential: Considering Liquidity and Solvency Interactions and Systemic Risk.” BCBS Working Paper 29, Bank for Interna-tional Settlements, Basel, Switzerland, November. https://www .bis.org/bcbs/publ/wp29.htm.

———. 2016. “Literature Review on Integration of Regulatory Capital and Liquidity Instruments.” BCBS Working Paper 30, Bank for International Settlements, Basel, Switzerland, March. http://www.bis.org/bcbs/publ/wp30.htm.

———. 2017a. “Basel III—The Net Stable Funding Ratio: Fre-quently Asked Questions.” BCBS Publication 396, Bank for International Settlements, Basel, Switzerland, February. https://www.bis.org/bcbs/publ/d396.htm.

———. 2017b. “Basel III—The Liquidity Coverage Ratio: Fre-quently Asked Questions.” BCBS Publication 406, Bank for International Settlements, Basel, Switzerland, October. https://www.bis.org/bcbs/publ/d406.htm.

———. 2017c. “Basel III: Finalising Post-crisis Regulatory Re-form.” BCBS Publication 424, Bank for International Settle-ments, Basel, Switzerland, April. https://www.bis.org/bcbs /publ/d424.htm.

———. 2017d. “Stress Testing Principles.” BCBS Publication 428, Bank for International Settlements, Basel, Switzerland, Decem-ber. https://www.bis.org/bcbs/publ/d428.htm.

———. 2018. “Fifteenth Progress Report on Adoption of the Basel Regulatory Framework.” BCBS Publication 452, Bank for Inter-national Settlements, Basel, Switzerland, April. https://www.bis .org/bcbs/publ/d452.htm.

Bucalossi, Annalisa, Cristina Fonseca Coutinho, Kerstin Junius, Alaoishe Luskin, Angeliki Momtsia, Imene Rahmouni-Rous-seau, Benjamin Sahel, Antonio Scalia, Stefan Schmitz, Franziska Schobert, Rita I. Soares, and Michael Wedow. 2016. “Basel III and Recourse to Eurosystem Monetary Policy Operations.” Oc-casional Paper 171, European Central Bank, Frankfurt am Main, Germany, April. https://www.ecb.europa.eu/pub/research /occasional-papers/html/papers-2016.en.html.

Caceres, Carlos, Christiana Daniel, Christoph Aymanns, and Lili-ana Schumacher. 2016. “Bank Solvency and Funding Costs.” IMF Working Paper 16/64, International Monetary Fund,

REFERENCESAikman, David, Piergiorgio Alessandri, Bruno Eklund, Prasanna Gai,

Sujit Kapadia, Elizabeth, Martin, Nada Mora, Gabriel Sterne, and Matthew Willison. 2009. “Funding Liquidity Risk in a Quantita-tive Model of Systemic Liquidity.” Bank of England Working Paper 372, Bank of England, London, UK, June. http://www .bankofengland.co.uk/publications/Pages/workingpapers/2009 /wp372.aspx.

Aiyar, Shekhar, Wolfgang Bergthaler, Jose Garrido, Anna Ilyina, Andreas A. Jobst, Kenneth Kang, Dmitriy Kovtun, Yan Liu, Dermot Monaghan, and Marina Moretti. 2015. “A Strategy for Resolving Europe’s Bad Loans.” IMF Staff Discussion Note 15/19, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/Staff-Discussion-Notes/Issues /2016/12/31/A-Strategy-for-Resolving-Europe-s-Problem-Loans -43286.

Anand, Kartik, Guillaume Bédard-Pagé, and Virginie Traclet. 2014. “Stress Testing the Canadian Banking System: A Sys-tem-Wide Approach.” Financial System Review (June): 61–68. https://www.bankofcanada.ca/2014/06/fsr-june-2014/.

Annaert, Jan, Marc De Ceustera, Patrick Van Roy, and Cristina Vespro. 2013. “What Determines Euro Area Bank CDS Spreads?” Journal of International Money and Finance 32: 444–61.

Barnhill, Theodore, and Liliana Schumacher. 2011. “Modeling Cor-related Systemic Liquidity and Solvency Risks in a Financial En-vironment with Incomplete Information.” IMF Working Paper 11/263, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Modeling -Correlated-Systemic-Liquidity-and-Solvency-Risks-in-a -Financial-Environment-with-25356.

Basel Committee on Banking Supervision (BCBS). 2008a. “Li-quidity Risk: Management and Supervisory Challenges.” BCBS Publication 136, Bank for International Settlements, Basel, Switzerland, February. https://www.bis.org/publ/bcbs136.htm.

———. 2008b. “Principles for Sound Liquidity Risk Management and Supervision.” BCBS Publication 144, Bank for Interna-tional Settlements, Basel, Switzerland, September. https://www .bis.org/publ/bcbs144.htm.

———. 2009. “Principles for Sound Stress Testing Practices and Supervision.” BCBS Publication 155, Bank for International Settlements, Basel, Switzerland, May. https://www.bis.org /publ/bcbs155.htm.

———. 2010a. “Results of the Comprehensive Quantitative Impact Study.” BCBS Publication 186, Bank for International Settle-ments, Basel, Switzerland, December. https://www.bis.org/publ /bcbs186.htm.

———. 2010b. “Basel III: International Framework for Liquidity Risk Measurement, Standards and Monitoring.” BCBS Publi-cation 188, Bank for International Settlements, Basel, Switzer-land, December. https://www.bis.org/publ/bcbs188.htm.

———. 2011. “Basel III: A Global Regulatory Framework for More Resilient Banks and Banking Systems.” BCBS Publica-tion 189, Bank for International Settlements, Basel, Switzer-land, June. https://www.bis.org/publ/bcbs189.htm.

———. 2012a. “Peer Review of Supervisory Authorities’ Imple-mentation of Stress Testing Principles.” BCBS Publication 218, Bank for International Settlements, Basel, Switzerland, April. https://www.bis.org/publ/bcbs218.htm.

———. 2012b. “Basel III Liquidity Standard and Strategy for As-sessing Implementation of Standards Endorsed by Group of














http://www.bis.org/bcbs/publ/wp30.htm.










https://www.ecb.europa.eu/pub/research/occasional-papers/html/papers-2016.en.html

https://www.ecb.europa.eu/pub/research/occasional-papers/html/papers-2016.en.html

http://www.bankofengland.co.uk/publications/Pages/workingpapers/2009/wp372.aspx



https://www.imf.org/en/Publications/Staff-Discussion-Notes/Issues/2016/12/31/A-Strategy-for-Resolving-Europe-s-Problem-Loans-43286




https://www.bankofcanada.ca/2014/06/fsr-june-2014/

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Modeling-Correlated-Systemic-Liquidity-and-Solvency-Risks-in-a-Financial-Environment-with-25356















Tests?” IMF Working Paper 14/103, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/How-to-Capture-Macro-Financial -Spillover-Effects-in-Stress-Tests-41644.

International Monetary Fund (IMF). 2008. Global Financial Sta-bility Report—Containing Systemic Risks and Restoring Financial Soundness, Chapter 3. Washington, DC, April. https://www .imf.org/en/Publications/GFSR/Issues/2016/12/31/Containing -Systemic-Risks-and-Restoring-Financial-Soundness.

———. 2010a. Global Financial Stability Report—Sovereigns, Funding, and Systemic Liquidity, Chapter 2. Washington, DC, October. https://www.imf.org/external/pubs/ft/gfsr/2010/02/.

———. 2010b. “Integrating Stability Assessments under the Finan-cial Sector Assessment Program into Article IV Surveillance.” IMF Policy Paper, Washington, DC, August. https://www.imf .org/en/Publications/Policy-Papers/Issues/2016/12/31/Integrating-Stability-Assessments-Under-the-Financial-Sector- Assessment-Program-into-PP4477.

———. 2010c. “United States: Stress Testing—Technical Note.” IMF Country Report 10/244, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/United-States -Publicat ion-of-Financia l-Sector-Assessment-Program -Documentation-Technical-24101.

———. 2010d. “Indonesia: Financial System Stability Assess-ment.” IMF Country Report 10/288, Washington, DC, June. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /Indonesia-Financial-System-Stability-Assessment-24212.

———. 2010e. “Finland: Financial System Stability Assessment— Update.” IMF Country Report 10/275, Washington, DC, Septem-ber. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /Finland-Financial-System-Stability-Assessment-Update-24182.

———. 2011a. Global Financial Stability Report—Durable Financial Stability: Getting There from Here, Chapter 2. Washington, DC, April. https://www.imf.org/en/Publications/GFSR/Issues/ 2016/12/31/Durable-Financial-Stability-Getting-There-from -Here.

———. 2011b. “Netherlands: Financial System Stability Assess-ment.” IMF Country Report 11/144, Washington, DC, June. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /Kingdom-of-Netherlands-Netherlands-Financial-System -Stability-Assessment-24986.

———. 2011c. “Luxembourg: Financial System Stability Assess-ment—Update.” IMF Country Report 11/148, Washington, DC, June. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Luxembourg-Financial-System-Stability-Assessment -Update-24995.

———. 2011d. “United Kingdom FSAP Update: Stress Testing the Banking Sector—Technical Note. IMF Country Report 11/227, Washington, DC, July. https://www.imf.org/en /Publications/CR/Issues/2016/12/31/United-Kingdom-Stress -Testing-the-Banking-Sector-Technical-Note-25119.

———. 2011e. “Sweden: Stress Testing of the Banking Sector—Technical Note.” IMF Country Report 11/288, Washington, DC, September. https://www.imf.org/en/Publications/CR/Issues /2016/12 /31/Swe den-Fi n a nc i a l - S e c tor-A s s e s sment -Program-Update-Technical-Note-on-Stress-Testing-of-the -25243.

———. 2011f. “People’s Republic of China: Financial System Sta-bility Assessment.” IMF Country Report 11/321, Washington, DC, November. https://www.imf.org/en/Publications/CR/Issues /2016/12/31/People-s-Republic-of-China-Financial-System -Stability-Assessment-25350.

Washington, DC. https://www.imf.org/en/Publications/WP /Issues/2016/12/31/Bank-Solvency-and-Funding-Cost-43792.

Catalán, Mario. 2015. “Treatment of Liquidity Risk in Stress Tests.” Unpublished Guidance Note on Stress Testing,” Mone-tary and Capital Markets Department, International Monetary Fund, Washington, DC.

Cerutti, Eugenio, Anna Ilyina, Yulia Makarova, and Christian Schmieder. 2010. “Bankers without Borders? Implications of Ring-Fencing for European Cross-Border Bank.” Working Paper 10/247, International Monetary Fund, Washington, DC. https://w w w.imf.org/en/Publ icat ions/W P/Issues/2016/12/31 /Bankers-Without-Borders-Implications-of-Ring-Fencing-for -European-Cross-Border-Banks-24335.

Chailloux, Alexandre, and Andreas A. Jobst. 2010. “Systemic Li-quidity Risk: Improving the Resilience of Financial Institutions and Markets, Box 2.5, Repo Infrastructure: Trading, Clearing, and Settlement.” In Global Financial Stability Report. Interna-tional Monetary Fund, Washington, DC, October. https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Global -Financial-Stability-Report-October-2010-Sovereigns-Funding- and-Systemic-Liquidity-23543.

Čihák, Martin. 2007. “Introduction to Applied Stress Testing.” IMF Working Paper 07/59, International Monetary Fund, Washington, DC, March. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/Introduction-to-Applied-Stress-Testing-20222.

Committee on the Global Financial System (CGFS). 2011. “Global Liquidity—Concept, Measurement and Policy Implications.” CGFS Paper 45, Bank for International Settlements, Basel, Switzerland, November. http://www.bis.org/publ/cgfs45.htm.

———. 2016. “Fixed Income Market Liquidity.” CGFS Paper 55, Bank for International Settlements, Basel, Switzerland, Janu-ary. https://www.bis.org/publ/cgfs55.htm.

Coval, Joshua, and Erik Stafford. 2007. “Asset Fire Sales (and Pur-chases) in Equity Markets.” Journal of Financial Economics 86 (2): 479–512.

Espinosa-Vega, Marco, and Juan Solé. 2011. “Cross-Border Finan-cial Surveillance: A Network Perspective.” Journal of Financial Economic Policy 3 (3): 82−205.

European Central Bank (ECB). 2008. “EU Banks Liquidity Stress Tests and Contingency Funding Plan.” European Central Bank, Frankfurt am Main, Germany. https://publications . europa.eu/en/publication-detail/-/publication/4fe7e351-b3b2 -453d-ae85-4fa4d5684922/language-en.

European Systemic Risk Board (ESRB). 2012. “Recommendation of the European Systemic Risk Board on Funding of Credit In-stitutions.” ESRB/2012/2, European Central Bank, Frankfurt am Main, Germany, December. https://www.esrb.europa.eu /mppa/recommendations/html/index.en.html.

Financial Stability Board (FSB). 2012. “Strengthening Oversight and Regulation of Shadow Banking: An Integrated Overview of Policy Recommendations.” Bank for International Settle-ments, Basel, Switzerland, November. http://www.fsb.org/2012 /11/r_121118/.

Gray, Dale, Laura Valderrama, Mario Catalán, and TengTeng Xu. 2017. “Macro-Financial Feedbacks in Stress Testing.” Presenta-tion at Joint IMF-EBA Colloquium—New Frontiers on Stress Testing, London, March 1.

Hasan, Iftekhar, Liu, Liuling, and Gaiyan Zhang. 2016. “The De-terminants of Global Bank Credit-Default-Swap Spreads.” Journal of Financial Services Research 50 (3): 275–309.

Hesse, Heiko, Ferhan Salman, and Christian Schmieder. 2014. “How to Capture Macro-Financial Spillover Effects in Stress


https://www.imf.org/en/Publications/WP/Issues/2016/12/31/How-to-Capture-Macro-Financial-Spillover-Effects-in-Stress-Tests-41644



https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Containing-Systemic-Risks-and-Restoring-Financial-Soundness.












https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Indonesia-Financial-System-Stability-Assessment-24212.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Indonesia-Financial-System-Stability-Assessment-24212.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Finland-Financial-System-Stability-Assessment-Update-24182.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Finland-Financial-System-Stability-Assessment-Update-24182.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Kingdom-of-Netherlands-Netherlands-Financial-System-Stability-Assessment-24986.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Luxembourg-Financial-System-Stability-Assessment-Update-24995.










https://www.imf.org/en/Publications/CR/Issues/2016/12/31/People-s-Republic-of-China-Financial-System-Stability-Assessment-25350



https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Bank-Solvency-and-Funding-Cost-43792

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Bank-Solvency-and-Funding-Cost-43792





https://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Global-Financial-Stability-Report-October-2010-Sovereigns-Funding-and-Systemic-Liquidity-23543




https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Introduction-to-Applied-Stress-Testing-20222

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Introduction-to-Applied-Stress-Testing-20222

http://www.bis.org/publ/cgfs45.htm.

https://www.bis.org/publ/cgfs55.htm

https://publications


https://www.esrb.europa.eu/mppa/recommendations/html/index.en.html

https://www.esrb.europa.eu/mppa/recommendations/html/index.en.html

http://www.fsb.org/2012/11/r_121118/

http://www.fsb.org/2012/11/r_121118/



June. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /Brazi l-Technica l-Note-on-Stress-Testing-the-Banking -Sector-40589.

———. 2013f. “France: Stress Testing the Banking Sector—Tech-nical Note.” IMF Country Report 13/185, Washington, DC, June. https://www.imf.org/en/Publications/CR/Issues/2016/12 /31/ Fr a nc e -F i n a nc i a l - S e c to r-A s s e s sment-Prog r a m -Technical-Note-on-Stress-Testing-the-Banking-40722.

———. 2013g. “Poland: Financial System Stability Assessment.” IMF Country Report 13/221, Washington, DC, July. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Republic -of-Poland-Financial-System-Stability-Assessment-40809.

———. 2013h. “Italy: Stress Testing the Banking Sector—Technical Note.” IMF Country Report 13/349, Washington, DC, December. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Italy-Technical-Note-on-Stress-Testing-The-Banking -Sector-41090.

———. 2013i. “Singapore: Financial System Stability Assess-ment.” IMF Country Report 13/325, Washington, DC, No-vember. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Singapore-Financial-System-Stability-Assessment -41051.

———. 2014a. “IMF Executive Board Reviews Mandatory Finan-cial Stability Assessments under the Financial Sector Assess-ment Program.” IMF Press Release 14/08, Washington, DC, January. https://www.imf.org/en/News/Articles/2015/09/14/01 /49/pr1408.

———. 2014b. “Austria: Stress Testing the Banking Sector—Techni-cal Note.” IMF Country Report 14/16, Washington, DC, Janu-ary. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /Aust r ia-Publ icat ion-of-Financia l-Sector-A sse ssment -Program-Documentation-Technical-Note-on-41266.

———. 2014c. “Canada: Stress Testing—Technical Note.” IMF Country Report 14/69, Washington, DC, March. https://w w w. i m f .org /en / Publ ic a t ion s /CR / I s sue s /2016/12/ 31/Canada-Financial-Sector-Assessment-Program-Stress -Testing-Technical-Note-41405.

———. 2014d. “People’s Republic of China–Hong Kong Special Administrative Region: Stress Testing the Banking Sector—Technical Note.” IMF Country Report 14/210, Washington, DC, July. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Peoples-Republic-of-ChinaHong-Kong-Specia l -Administrative-Region-Financial-Sector-Assessment-41755.

———. 2014e. “Switzerland: Stress Testing the Banking System—Technical Note.” IMF Country Report 14/267, Washington, DC, September. https://www.imf.org/en/Publications/CR/Issues /2016/12/31/Switzerland-Technical-Note-Stress-Testing-the -Banking-System-41883.

———. 2014f. “Denmark: Stress Testing the Banking, Insurance, and Pension Sectors—Technical Note.” IMF Country Report 14/348, Washington, DC, December. https://www.imf.org/en /Publications/CR/Issues/2016/12/31/Denmark-Stress-Testing -the-Banking-Insurance-and-Pension-Sectors-Technical -Note-42536.

———. 2015a. “Korea: Stress Testing and Financial Stability Analysis—Technical Note.” IMF Country Report 15/6, Wash-ington, DC, January. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Republic-of-Korea-Financial-Sector -Assessment-Program-Stress-Testing-And-Financial-Stability -42585.

———. 2015b. “South Africa: Stress Testing the Financial Sys-tem—Technical Note.” IMF Country Report 15/54, Washing-

———. 2011g. “Russian Federation: Stress Testing—Technical Note.” IMF Country Report 11/334, Washington, DC, De-cember. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Russian-Federat ion-Technica l-Note-on-Stress -Testing-of-the-Banking-Sector-25385.

———. 2011h. “Germany: Stress Testing—Technical Note.” IMF Country Report 11/371, Washington, DC, December. https://w w w.imf.org /en/Publ icat ions/CR /Issues/2016/12/31 /Germany-Technical-Note-on-Stress-Testing-25461.

———. 2012a. “Mexico: Financial System Stability Assessment.” IMF Country Report 12/65, Washington, DC, June. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Mexico -Financial-System-Stability-Assessment-25810.

———. 2012b. “Spain: Financial System Stability Assessment.” IMF Country Report 12/137, Washington, DC, June. https://w w w.imf.org /en/Publ icat ions/CR /Issues/2016/12/31 /Spain-Financial-System-Stability-Assessment-25977.

———. 2012c. “Japan: Financial System Stability Assessment—Update.” IMF Country Report 12/210, Washington, DC, Au-gust. https://www.imf.org/en/Publications/CR/Issues/2016/12 /31/Japan-Financial-Sector-Stability-Assessment-Update -26137.

———. 2012d. “Turkey: Financial System Stability Assessment—Update.” IMF Country Report 12/261, Washington, DC, Sep-tember. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Turkey-Financial-System-Stability-Assessment-26245.

———. 2012e. “Macro-financial Stress Testing—Principles and Practices.” IMF Policy Paper, Washington, DC, August. https://www.imf.org/en/Publications/Policy-Papers/Issues /2016/12/31/Macrofinancial-Stress-Testing-Principles-and- Practices-PP4702.

———. 2012f. “Saudi Arabia: Financial System Stability Assess-ment—Update.” IMF Country Report 12/92, Washington, DC, April. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Saudi-Arabia-Financial-System-Stability-Assessment -Update-25859.

———. 2012g. “Australia: Financial System Stability Assessment.” IMF Country Report 12/308, Washington, DC, November. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 /Australia-Financial-System-Stability-Assessment-40107.

———. 2013a. “Mandatory Financial Stability Assessments under the Financial Sector Assessment Program: Update.” IMF Policy Paper, Washington, DC, November. https://www.imf.org/ e n / P ub l i c a t i on s / Po l i c y-Pa p e r s / I s s u e s /2 016/12 /31 /Mandatory-Financia l-Stability-Assessments-Under-the -Financial-Sector-Assessment-Program-PP4838.

———. 2013b. “European Union: Stress Testing of Banks—Tech-nical Note.” IMF Country Report 13/68, Washington, DC, March. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/European-Union-Publication-of-Financial-Sector -Assessment-Program-Documentation-Technical-40396.

———. 2013c. “India: Financial System Stability Assessment.” IMF Country Report 13/8, Washington, DC, January. http://w w w.imf.org /en/Publ icat ions/CR /Issues/2016/12/31 /India-Financial-System-Stability-Assessment-Update-40231.

———. 2013d. “Belgium: Stress Testing the Banking and Insur-ance Sectors—Technical Note.” IMF Country Report 13/137, Washington, DC, May. https://www.imf.org/en/Publications /CR/Issues/2016/12/31/Belgium-Technical-Note-on-Stress -Testing-the-Banking-and-Insurance-Sectors-40573.

———.2013e. “Brazil: Stress Testing the Banking Sector—Tech-nical Note.” IMF Country Report 13/147, Washington, DC,


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Brazil-Technical-Note-on-Stress-Testing-the-Banking-Sector-40589.






https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Republic-of-Poland-Financial-System-Stability-Assessment-40809.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Italy-Technical-Note-on-Stress-Testing-The-Banking-Sector-41090.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Singapore-Financial-System-Stability-Assessment-41051.



https://www.imf.org/en/News/Articles/2015/09/14/01/49/pr1408

https://www.imf.org/en/News/Articles/2015/09/14/01/49/pr1408

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Austria-Publication-of-Financial-Sector-Assessment-Program-Documentation-Technical-Note-on-41266.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Canada-Financial-Sector-Assessment-Program-Stress-Testing-Technical-Note-41405.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Peoples-Republic-of-ChinaHong-Kong-Special-Administrative-Region-Financial-Sector-Assessment-41755.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Switzerland-Technical-Note-Stress-Testing-the-Banking-System-41883.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Denmark-Stress-Testing-the-Banking-Insurance-and-Pension-Sectors-Technical-Note-42536.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Republic-of-Korea-Financial-Sector-Assessment-Program-Stress-Testing-And-Financial-Stability-42585.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Russian-Federation-Technical-Note-on-Stress-Testing-of-the-Banking-Sector-25385.






https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Mexico-Financial-System-Stability-Assessment-25810.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Spain-Financial-System-Stability-Assessment-25977.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Japan-Financial-Sector-Stability-Assessment-Update-26137.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Turkey-Financial-System-Stability-Assessment-26245.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Turkey-Financial-System-Stability-Assessment-26245.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Saudi-Arabia-Financial-System-Stability-Assessment-Update-25859.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Australia-Financial-System-Stability-Assessment-40107.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Australia-Financial-System-Stability-Assessment-40107.

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Mandatory-Financial-Stability-Assessments-Under-the-Financial-Sector-Assessment-Program-PP4838.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/European-Union-Publication-of-Financial-Sector-Assessment-Program-Documentation-Technical-40396.



http://www.imf.org/en/Publications/CR/Issues/2016/12/31/India-Financial-System-Stability-Assessment-Update-40231.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Belgium-Technical-Note-on-Stress-Testing-the-Banking-and-Insurance-Sectors-40573..2013e





———. 2018. “The Financial Sector Assessment Program— Factsheet.” International Monetary Fund, Washington, DC, March 8. https://www.imf.org/en/About/Factsheets/Sheets/2016 /08/01/16/14/Financial-Sector-Assessment-Program.

Jobst, Andreas A. 2011. “Systemic Liquidity Risk and Macropru-dential Stress Testing,” Presentation, Stream 2 – Liquidity Risk Modelling and Management, ALM Europe/Basel III Congress, RISK Magazine, 28 September (London).

———. 2012. “Measuring Systemic Risk-Adjusted Liquidity (SRL)—A Model Approach.” IMF Working Paper 12/209, In-ternational Monetary Fund, Washington, DC. https://www.imf .org/en/Publications/WP/Issues/2016/12/31/Measuring -Systemic-Risk-Adjusted-Liquidity-SRL-A-Model-Approach -26203.

———. 2014. “Measuring Systemic Risk-Adjusted Liquidity (SRL)—A Model Approach.” Journal of Banking and Finance 45: 270–87.

———. 2017. “Top-Down Bank Liquidity Stress Testing Tool.” Attachment to IMF Working Paper 17/102, “Macroprudential Liquidity Stress Testing in FSAPs for Systemically Important Financial Systems.” https://www.imf.org/en/Publications/WP /Issues/2017/05/01/Macroprudential-Liquidity-Stress-Testing -in-FSAPs-for-Systemically-Important-Financial-44873 .

Jobst, Andreas A, Ong, Li Lian, and Christian Schmieder. 2013. “A Framework for Macroprudential Bank Solvency Stress Test-ing: Application to S-25 and Other G20 Country FSAPs.” IMF Working Paper 13/68, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/A-Framework-for-Macroprudential-Bank-Solvency -Stress-Testing-Application-to-S-25-and-Other-G-40390.

Jobst, Andreas A, Li Lian Ong, and Christian Schmieder. 2017. “Macroprudential Liquidity Stress Testing in FSAPs for Sys-temically Important Financial Systems.” IMF Working Paper 17/102, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP/Issues/2017/05/01 /Macroprudential-Liquidity-Stress-Testing-in-FSAPs-for -Systemically-Important-Financial-44873.

Krznar, Ivo, and Troy Matheson. 2017. “Towards Macroprudential Stress Testing: Incorporating Macro Feedback Effects.” IMF Working Paper 17/149, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues/2017 /06/30/Towards-Macroprudential-Stress-Testing-Incorporating -Macro-Feedback-Effects-44955.

Markets Committee. 2016. “Electronic Trading in Fixed Income Markets.” Bank for International Settlements, Basel, January. https://www.bis.org/publ/mktc07.htm.

Nier, Erlend, Nicolas Arregui, Chikako Baba, R. Sean Craig, Gavin Gray, Heedon Kang, Ivo Krznar, and Nadege Jassaud. 2014. “Staff Guidance Note on Macroprudential Policy— Detailed Guidance on Instruments.” International Monetary Fund, Washington, DC, December. https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Staff-Guidance -Note-on-Macroprudential-Policy-Detailed-Guidance-on -Instruments-PP4928.

Österreichische Nationalbank (OeNB). 2009. Financial Stability Report 18. Vienna, Austria: Oesterreichische Nationalbank. https://www.oenb.at/en/Publications/Financial-Market/Financial -Stability-Report/2009/Financial-Stability-Report-18.html.

Ong, Li Lian, and Martin Čihák. 2010. “Of Runes and Sagas: Per-spective on Liquidity Stress Testing Using an Iceland Example.” IMF Working Paper 10/156, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications/WP

ton, DC, March. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/South-Africa-Financial-Sector-Assessment -Program-Stress-Testing-the-Financial-System-42756.

———. 2015c. “United States: Stress Testing—Technical Note.” IMF Country Report 15/173, Washington, DC, July. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United -States-Financial-Sector-Assessment-Program-Stress-Testing -Technical-Notes-43058.

———. 2015d. “Norway: Stress Testing the Banking Sector—Technical Note.” IMF Country Report 15/258, Washington, DC, September. https://www.imf.org/en/Publications/CR/Issues /2016/12/31/Norway-Financial-Sector-Assessment-Program -Technical-Note-Stress-Testing-the-Banking-Sector-43271.

———. 2015e. “The Financial Sector Assessment Program— Factsheet.” International Monetary Fund, Washington, DC, September 21.

———. 2015f. Global Financial Stability Report—Vulnerabilities, Legacies, and Policy Challenges, Chapter 2. Washington, DC, October. http://www.imf.org/en/Publications/GFSR/Issues/2016 /12/31/Vulnerabilities-Legacies-and-Policy-Challenges.

———. 2016a. “Argentina: Financial Sector Stability—Technical Note.” IMF Country Report 16/65, Washington, DC, Febru-ary. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Argentina-Financial-Sector-Assessment-Program-Financial -Sector-Stability-Technical-Note-43739.

———. 2016b. “United Kingdom: Stress Testing the Banking Sector—Technical Note.” IMF Country Report 16/163, Wash-ington, DC, June 17. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-Kingdom-Financial-Sector -Assessment-Program-Stress-Testing-the-Banking-Sector -43974.

———. 2016c. “Germany: Stress Testing the Banking and Insur-ance Sectors—Technical Note.” IMF Country Report 16/191, Washington, DC, June 29. https://www.imf.org/en/Publications /CR/Issues/2016/12/31/Germany-Financial-Sector-Assessment -Program-Stress-Testing-the-Banking-and-Insurance-Sectors -44015.

———. 2016d. “Russian Federation: Technical Note Stress Test-ing.” IMF Country Report 16/306, Washington, DC, Septem-ber. https://www.imf.org/en/Publications/CR/Issues/2016/12 /31/Russian-Federat ion-Financia l-Sector-A ssessment -Program-Technical-Note-Stress-Testing-44288.

———. 2016e. “Ireland: Stress Testing the Banking System—Technical Note.” IMF Country Report 16/315, Washington, DC, September. http://www.imf.org/en/Publications/CR/Issues /2016/12 /31/ I r e l a nd-Fi n a nc i a l - S e c tor-A s s e s sment -Program-Technical-Note-Stress-Testing-the-Banking-System -44308.

———. 2017a. “Finland: Stress Testing the Banking System and Interconnectedness Analysis—Technical Note.” IMF Country Report 17/6, Washington, DC, January. https://www.imf.org /en/Publications/CR/Issues/2017/01/11/Finland-Financial -Sec tor-A s se s sment-Prog ra m-Techn ic a l-Note -St re s s -Testing-the-Banking-System-44516.

——— . 2017b. “Turkey: Financial System Stability Assessment.” IMF Country Report 17/35, Washington, DC, February. https://www.imf.org/en/Publications/CR/Issues/2017/02/03 /Turkey-Financial-Sector-Assessment-Program-Financial -System-Stability-Assessment-44617.

———. 2017c. “Financial Sector Assessment Program (FSAP).” International Monetary Fund, Washington, DC, December. http://www.imf.org/external/np/fsap/fssa.aspx.


https://www.imf.org/en/About/Factsheets/Sheets/2016/08/01/16/14/Financial-Sector-Assessment-Program

https://www.imf.org/en/About/Factsheets/Sheets/2016/08/01/16/14/Financial-Sector-Assessment-Program

https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Measuring-Systemic-Risk-Adjusted-Liquidity-SRL-A-Model-Approach-26203.




https://www.imf.org/en/Publications/WP/Issues/2017/05/01/Macroprudential-Liquidity-Stress-Testing-in-FSAPs-for-Systemically-Important-Financial-44873



https://www.imf.org/en/Publications/WP/Issues/2016/12/31/A-Framework-for-Macroprudential-Bank-Solvency-Stress-Testing-Application-to-S-25-and-Other-G-40390









https://www.bis.org/publ/mktc07.htm

https://www.imf.org/en/Publications/Policy-Papers/Issues/2016/12/31/Staff-Guidance-Note-on-Macroprudential-Policy-Detailed-Guidance-on-Instruments-PP4928




https://www.oenb.at/en/Publications/Financial-Market/Financial-Stability-Report/2009/Financial-Stability-Report-18.html

https://www.oenb.at/en/Publications/Financial-Market/Financial-Stability-Report/2009/Financial-Stability-Report-18.html


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/South-Africa-Financial-Sector-Assessment-Program-Stress-Testing-the-Financial-System-42756.



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-States-Financial-Sector-Assessment-Program-Stress-Testing-Technical-Notes-43058.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Norway-Financial-Sector-Assessment-Program-Technical-Note-Stress-Testing-the-Banking-Sector-43271.



http://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Vulnerabilities-Legacies-and-Policy-Challenges.

http://www.imf.org/en/Publications/GFSR/Issues/2016/12/31/Vulnerabilities-Legacies-and-Policy-Challenges.

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Argentina-Financial-Sector-Assessment-Program-Financial-Sector-Stability-Technical-Note-43739.







https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Germany-Financial-Sector-Assessment-Program-Stress-Testing-the-Banking-and-Insurance-Sectors-44015.




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Russian-Federation-Financial-Sector-Assessment-Program-Technical-Note-Stress-Testing-44288.



http://www.imf.org/en/Publications/CR/Issues/2016/12/31/Ireland-Financial-Sector-Assessment-Program-Technical-Note-Stress-Testing-the-Banking-System-44308.




https://www.imf.org/en/Publications/CR/Issues/2017/01/11/Finland-Financial-Sector-Assessment-Program-Technical-Note-Stress-Testing-the-Banking-System-44516




https://www.imf.org/en/Publications/CR/Issues/2017/02/03/Turkey-Financial-Sector-Assessment-Program-Financial-System-Stability-Assessment-44617.



http://www.imf.org/external/np/fsap/fssa.aspx


Schmitz, Stefan W., Michael Sigmund, and Laura Valderrama. 2017. “Bank Solvency and Funding Cost: New Data and New Results.” IMF Working Paper 17/116, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2017/05/15/Bank-Solvency-and-Funding-Cost -New-Data-and-New-Results-44914.

Schuermann, Til. 2012. “Stress Testing Banks.” Wharton Finan-cial Institutions Center Working Paper 12/08, University of Pennsylvania, Philadelphia. https://fic.wharton.upenn.edu/working -papers/archived-working-papers/2012-working-papers/.

Valderrama, Laura. 2016. “An Agent-Based Model for Stress Test-ing.” Unpublished Working Paper, March 20, International Monetary Fund, Washington, DC.

Van den End, Jan Willem. 2008. “Liquidity Stress-Tester: A Macro Model for Stress Testing Banks’ Liquidity Risk.” DNB Work-ing Paper 175, De Nederlandsche Bank Amsterdam, Nether-lands, May. https://www.dnb.nl/en/news/dnb-publications/dnb -working-papers-series/dnb-working-papers/auto175528.jsp.

Wong, Eric, and Cho-Hoi Hui. 2009. “A Liquidity Risk Stress Testing Framework with Interaction between Market and Credit Risks.” Working Paper 06/2009, Hong Kong Monetary Authority, Hong Kong, Special Administrative Region. https://www.hkma.gov.hk/eng/publications-and-research/research /working-papers/2009/.

/Issues/2016/12/31/Of-Runes-and-Sagas-Perspectives-on -Liquidity-Stress-Testing-Using-an-Iceland-Example-24019.

Pengelly, Mark. 2012. “Bail-in Plans Could Undermine Basel’s NSFR.” RISK Magazine (27 March). http://www.risk.net/risk -magazine/news/2163763/bail-plans-undermine-basels-nsfr.

Puhr, Claus, and Stefan W. Schmitz. 2014. “A View From The Top—The Interaction between Solvency and Liquidity Stress.” Journal of Risk Management in Financial Institutions 7 (4): 38–51.

Shleifer, Andrei, and Robert Vishny. 2010. “Fire Sales in Finance and Macroeconomics.” NBER Working Paper 16642, National Bureau of Economic Research, Cambridge, MA. http://www .nber.org/papers/w16642.

Schmieder, Christian, Heiko Hesse, Benjamin Neudorfer, Claus Puhr, and Stefan W. Schmitz. 2012. “Next Generation Sys-tem-wide Liquidity Stress Testing.” IMF Working Paper 12/3, International Monetary Fund, Washington, DC. https://www .imf.org/external/pubs/cat/longres.aspx?sk=25509.0.

Schmitz, Stefan W. 2015. “Macroprudential Liquidity Stress Tests.” In Liquidity Risk Management and Supervision, edited by Iman van Lelyveld, Paul Hilbers, and Clemens Bonner, 237–64. Lon-don: Risk Books.

Schmitz, Stefan W., and Andreas Ittner. 2007. “Why Central Banks Should Look at Liquidity Risk.” Central Banking, 17 (4): 32–40.


https://www.imf.org/en/Publications/WP/Issues/2017/05/15/Bank-Solvency-and-Funding-Cost-New-Data-and-New-Results-44914



https://fic.wharton.upenn.edu/working-papers/archived-working-papers/2012-working-papers/

https://fic.wharton.upenn.edu/working-papers/archived-working-papers/2012-working-papers/

https://www.dnb.nl/en/news/dnb-publications/dnb-working-papers-series/dnb-working-papers/auto175528.jsp

https://www.dnb.nl/en/news/dnb-publications/dnb-working-papers-series/dnb-working-papers/auto175528.jsp

https://www.hkma.gov.hk/eng/publications-and-research/research/working-papers/2009/./Issues/2016/12/31/Of-Runes-and-Sagas-Perspectives-on-Liquidity-Stress-Testing-Using-an-Iceland-Example-24019





http://www.risk.net/risk-magazine/news/2163763/bail-plans-undermine-basels-nsfr

http://www.risk.net/risk-magazine/news/2163763/bail-plans-undermine-basels-nsfr





CHAPTER 17

Macroprudential Solvency Stress Testing of the Insurance Sector

ANDREAS A. JOBST • NOBUYASU SUGIMOTO • TIMO BROSZEIT

Over the last decade, the IMF has placed more emphasis on the role of the insurance industry in its bilateral and multilateral financial sector surveillance. This chapter reviews the current state of system- wide insurance solvency stress tests based on a comparative review of national

practices and the experiences from the IMF’s Financial Sector Assessment Program with the aim of providing practical guidance for the coherent and consistent implementation of such exercises. The chapter also offers recommendations on improving the current insurance stress testing ap-proaches and presentation of results.

are defined by adverse changes in risk factors corresponding to exceptional but plausible events with considerable adverse im-pact on financial stability (CGFS 2000 and 2005). They can be limited to one sector or provide a cross- sectoral perspective by capturing the interconnectedness of banks, insurers, and other market participants.

Stress testing has become a central aspect of the IMF’s fi-nancial sector surveillance, reflecting its important role within the regulatory frameworks and as part of financial supervision in IMF member countries. It is a key component of the Finan-cial Sector Assessment Program (FSAP), and is also used in Article IV consultations and crisis program work.1 Based on the IMF’s practical experience with system- wide stress testing after more than a decade of FSAPs (with a focus on the bank-ing sector), in 2012, the IMF staff proposed a set of “best prac-tice” principles for macro- financial stress testing (IMF 2012b). This chapter applies and extends these principles to the macro-prudential stress testing of the insurance sector.

The purpose of FSAP stress tests differs from that of super-visory stress testing exercises. FSAP stress testing approaches are designed for surveillance purposes, with a medium- term focus. They typically involve very severe but plausible stress

1. INTRODUCTIONA key lesson from the global financial crisis has been a greater focus on concepts to identify the buildup of financial risks. This has spawned risk- based analytical frameworks for financial stability analysis, including the examination of macro- financial linkages and the integration of advanced market- and risk- based tools for bilateral and multilateral surveillance. These developments have underscored the need for a coherent and consistent approach to stress testing.

Stress testing is a forward- looking technique that aims at measuring the sensitivity of a portfolio, an institution, or even an entire financial system to adverse events that have a small probability of occurrence but a significant economic impact if they were to occur. Well- formulated stress tests comprise dif-ferent methods, such as sensitivity and/or scenario analyses, to assess the overall capacity of individual firms or an entire sec-tor to absorb negative shocks to macro- financial conditions. In financial sector stability analysis, stress tests are aimed at identifying vulnerabilities to the impact of a rapid deteriora-tion in the operational and market environment affecting the overall risk profile from the financial system level down to the individual firm and portfolio levels. Such adverse conditions

This chapter is based on IMF Working Paper 14/133 (Jobst, Sugimoto, and Broszeit). Some parts of this chapter were completed when Andreas Jobst was Chief Econo-mist and Deputy Director (Supervision) of the Bermuda Monetary Authority (BMA) as well as Vice Chair of the Financial Stability Data Specialists Subcommit-tee (FSD) of the International Association of Insurance Supervisors (IAIS). The authors would like to thank Kalin Tintchev for research assistance, Michaela Erbenova, Jorge Chan- Lau, and Martin Čihák for insightful comments, as well as Marcela Gronychová (CNB), Stanislav Georgiev (BaFin, now factualarguments GmbH), Knut Schäffler (BaFin), Hideki Takeuchi (Japan FSA), Marc Radice (formerly FINMA, now Zurich Insurance), Markus Bachmann (FINMA), Dean Minot (PRA), John Hopman (NAIC), and Philipp Keller (Deloitte, now Quantica Capital AG) for their invaluable feedback and input to this chapter.1 Stress tests first emerged in the late 1990s and have been used since then by central banks, regulatory bodies, and international organizations, such as the IMF and

the World Bank, to proactively identify vulnerabilities and/or to determine specific risks for certain industry sectors or systemically relevant financial institutions.


Macroprudential Solvency Stress Testing of the Insurance Sector446

scenarios to assess the overall resilience of the financial sys-tem to the realization of selected risk drivers.2 While the stress test results have no immediate supervisory implica-tions, they are an essential input into broader analyses un-dertaken by the FSAP team, forming the basis for policy discussions on financial stability issues with the authorities. This is different from supervisory stress tests, which are aimed at identifying any potential capital shortfall from the likely economic impact of one or more adverse events for which management actions may be required.3 These stress tests also help validate internal (economic) capital models to substan-tiate the resilience of the firm to extreme shocks.

In the context of FSAPs, insurance stress testing has played only a secondary role relative to the analysis of the banking sector risks. Since 1999, 31 stress tests of the insurance sector have been included in a total of 314 FSAPs and standards as-sessments (as of the end of April 2018),4 while bank stress test-ing has become an integral part of every mission (Figures 17.1 and 17.2). This might be explained not only by the fact that insurers are considered less systemically relevant in many ju-risdictions but also by the unique conceptual challenges aris-ing from their different balance sheet structure and the lack of global solvency and valuation standards impacting the design and comparability of top- down (TD) stress testing. This has also resulted in a greater reliance on national solvency regimes for stress testing in bottom- up (BU) approaches, which are more resource- intensive exercises.

Recent developments in macroprudential surveillance, how-ever, warrant greater focus on identifying systemic risks af-fecting the insurance sector, including through stress testing. Even though traditional insurance activities did not contrib-ute to systemic risk during the financial crisis, the assessment methodology underpinning the designation of global systemi-cally important insurance companies ( G- SIIs) identifies some vulnerabilities from nontraditional and/or noninsurance (NTNI) activities (IAIS 2013c).5 Moreover, the linkages be-tween insurers, banks, and other financial institutions may

increase in the future as a result of postcrisis regulatory reforms and greater integration of financial services. This could change the transmission channels of risks affecting the solvency and liquidity conditions of insurance companies under stress.

This chapter reviews the current state of system- wide sol-vency stress testing of insurers and provides guidelines for the consistent implementation of such tests for macropru-dential surveillance purposes. The focus is on the assessment of capital adequacy under adverse financial conditions to support a comprehensive understanding of general vulnera-bilities to shocks. Based on the practical insights gained from relevant FSAPs and stress testing approaches used by supervisory authorities, the chapter identifies, similarly to Jobst, Ong, and Schmieder 2013 for the banking sector, best practices and methodologies for such stress testing with a conceptual treatment of potential spillover and contagion ef-fects from the interlinkages of insurance companies with other financial institutions.6 It also augments the banking focus on asset risks in current stress testing approaches with a discussion of underwriting risks that affect the liabilities of insurance firms. Specifically, the chapter (1) articulates the main characteristics of a stress testing framework and dem-onstrates its application in the IMF surveillance of insurers, (2) compares the actual implementation of various stress tests in a range of major country FSAPs based on a detailed cross- country stress testing matrix (Appendix Tables 17.1.3 and 17.1.4), and (3) discusses the general properties of stress testing approaches for insurance as guidance for readers seeking to develop their own macroprudential stress testing frameworks and for country authorities preparing for FSAPs.

2. OVERVIEW AND FRAMEWORK

Macroprudential Stress Testing for Insurance

In the wake of the global financial crisis, there has been an increased focus on macroprudential policy and surveillance (MPS) with a view toward enhancing the resilience of the financial sector to systemic risk.7 MPS comprises the 2 Additional severity in FSAP stress testing for insurance stems from the

longer forecast horizon than is usually applied by supervisors and the combination of risk factors affecting both assets and liabilities in a com-prehensive scenario- based framework.

3 This would also involve capital plans designed to return the relevant firm to a stable, sustainable position, including options to address capi-tal shortfalls through generating capital internally and externally (also including restricting dividends and variable remuneration). Supervisors would assess the appropriateness of insurers’ plans in terms of the ade-quacy of identified recovery plans and the supporting governance struc-ture (PRA 2013).

4 This also includes 18 FSAPs that were ongoing during the completion of this chapter, of which four include an insurance stress test.

5 In July 2013, the International Association of Insurance Supervisors (IAIS)—in coordination with the Financial Stability Board—published its final version of an initial assessment methodology for the identification of G- SIIs together with a draft proposal of policy measures for designated firms (IAIS 2012a, 2012d, 2013b, 2013c, 2015), including enhanced su-pervision, effective recovery and resolution, and capital requirements. The weighted indicator- based approach for G- SIIs is similar in concept to that used to identify global systemically important banks, but also introduces additional indicators that are germane to insurance activities.

6 Note that the tenor of the chapter is on system- wide stress testing with-out an in- depth discussion of the spillover effects of insurance compa-nies to the financial system at large. This coverage is consistent with the assessment of systemic risk in the insurance sector (IAIS 2011a). While banks are prone to contribute to systemic risk from individual failures that propagate material financial distress or activities via intrasectoral and intersectoral linkages to other institutions and markets (based on direct exposures via lending and investment), insurers tend to be more affected by their common exposures to asset price shocks that challenge the overall resilience of the sector (Jobst 2014). Hesse, Ferhan, and Schmieder (2014) provide general examples of modeling spillover ef-fects in macroprudential solvency stress tests for banking.

7 See IMF 2011c and 2013c for an overview of current theoretical and empirical work on macroprudential policy and regulation. The CGFS (2012) recently published a report on operationalizing macroprudential policies. See also IMF 2011c and 2011d for a more empirically focused review of macroprudential surveillance and its implementation for fi-nancial stability analysis.


Andreas A. Jobst, Nobuyasu Sugimoto, and Timo Broszeit 447

the investment and underwriting performances of the insur-ance sector within the broader financial system. More spe-cifically, MPS in the insurance sector comprises a three- step process (IAIS 2013a):

1. Determining key indicators of general macro- financial vulnerabilities of different insurance business models and recognizing the need to distinguish traditional and NTNI activities.

identification, measurement, and monitoring of the potential buildup of vulnerabilities of multiple firms within a country and/or across national boundaries with the goal of mitigating systemic risk. The scope of monitoring goes beyond institu-tional fragility, especially in areas of economic significance to both financial sector participants and the real economy.

Macroprudential stress testing for insurance builds on identifiable transmission channels of systemic risk affecting

Number of FSAPs/IMF insurance stress tests

0123IMF insurancestress tests

Source: Authors.Note: FSAP = Financial Sector Assessment Program.1Includes full FSAPs, FSAP Updates, and Stability Modules (including reviews under the Offshore Financial Centers Assessment Program). The overview does not show the insurance stress tests completed as part of the assessment for Bermuda, Guernsey, and the Isle of Man. Belgium, Denmark, France, Japan, Singapore, South Africa, and the United States are the only countries that have completed two FSAP insurance stress test exercises so far.

Figure 17.1 Overview of IMF FSAPs1 and Completion of Insurance Stress Tests (January 2000–April 2018)

Insurance STFSAPs without insurance sector ST

40

20

0

60

80

100

120

140

160

180

All countries(- 2008)

All countries(since 2009)

Advanced economies(- 2008)

Advanced economies(since 2009)

Num

ber o

f FSA

Ps1

13

18

1117

Source: IMF staff.Note: FSAP = Financial Sector Assessment Program; ST = stress test.1Total number comprises full FSAPs, FSAP Updates, and Stability Modules (including all countries subject to the Offshore Financial Centers Assessment Program with a stress testing exercise) but excludes three (out of a total of 30) insurance stress tests that have not been published.

Figure 17.2 Number of Completed Insurance Stress Tests in FSAPs1 before and after the Global Financial Crisis, as of End of December 2016



While bank stress tests consider the sector’s vulnerability to general deterioration in economic conditions (based on his-torical and/or market- implied sensitivities of all profit and loss components), most insurance stress tests tend to focus on the aggregate impact of very specific changes in macro- financial indicators, such as interest rates/credit spreads, asset risks, and foreign exchange rates. For underwriting risks, specific stress test approaches need to be applied. The assessment of vulnerabilities arising from existing liabilities in life and non- life insurance under stress is essential to a comprehensive assessment and comprises the deteriora-tion of technical provisions,9 demographic risks, and cata-strophic risks.10

Differences between Banks and Insurance Companies and Their Implications for Stress Testing11

The nature of risk taking of banks and insurance companies is markedly different and suggests limited usefulness of most common stress testing approaches, which tend to be focused on banking activities. While insurance companies share some similarities with banks, they do not engage in matu-rity/liquidity transformation as a key source of systemic risk when solvency and liquidity stresses coincide.12 Banks incur mostly short- term liabilities to finance longer term assets, such as commercial and retail loans, including mortgages. In contrast, insurers are funded by upfront premium payments, resulting in more stable cash flows (than are seen in the banking model).13 The liabilities of an insurance company (which, for a life insurer, would usually be of a long- term nature) are mostly technical provisions for insurance claims (Figure 17.3), which are backed by a diversified investment portfolio composed of mostly high- quality assets (whose cash flows are closely aligned with the expected claims expe-rience). For banks, where sharp asset price declines may lead to immediate and substantial liquidity drains and solvency challenges, insurers typically do not suffer collateral calls or

2. Designing a conceptual approach for defining risk fac-tors and transmission channels of stress aimed at identi-fying common exposures, risk concentrations, and interdependencies that are sources of spillover effects and contagion risks that may jeopardize the function-ing of the system as a whole.

3. Developing a macroprudential framework that integrates the key risk drivers of macro- financial vulnerabilities with the design and implementation of meaningful and relevant shocks to risk factors to determine super-visory action, operational changes, and/or suitable policy measures that can mitigate the severity and duration of material distress affecting the insurance sector (with adverse effects on the real economy).

However, the development of MPS in the insurance sec-tor is still in its infancy. An IAIS survey of macroprudential surveillance practices at the national level revealed that most supervisory authorities carry out macroprudential surveil-lance activities (IAIS 2010), but the use of insurance- specific stress testing for MPS is limited. The two most prevalent ap-proaches within MPS are (1) the monitoring of trends and development in insurance markets and (2) the analysis of the system- wide impact of macroeconomic variables on the in-surance market. In both instances, the focus tends to be on the analysis of domestic data, with international data analy-sis receiving comparatively less attention.8

Stress tests have increasingly been used by insurance super-visors, but more for microprudential purposes. When the IAIS (2011b) revised the Insurance Core Principles (ICPs), it also introduced enterprise risk management for solvency pur-poses, including stress testing and scenario analysis. Never-theless, such analysis remains focused on the viability of individual institutions after the economic impact of shocks rather than the system- wide robustness to the joint impact of risks arising from the (1) growing complexity of the intercon-nectedness among insurance companies and with other finan-cial institutions, and (2) the extent to which such interlinkages cause potential spillover and contagion effects. Although su-pervisors are not explicitly required to conduct system- wide stress testing, they are expected to monitor vulnerabilities within the insurance sector and carry out the analysis of “plausible unfavorable future scenarios with the objective and capacity to take action at an early stage, if required,” aimed at identifying and mitigating systemic risk that might negatively affect the risk profile of insurers (ICP 24). Stress tests have also been increasingly used to analyze market dynamics under ex-treme (tail risk) scenarios in order to ascertain whether or not precautionary supervisory intervention would be warranted.

Solvency stress tests for insurance tend to assess the capi-tal impact of shocks to risk factors on the total balance sheet.

8 In 2013, the IAIS (IAIS 2013a) published the first guidelines of MPS for insurance. Similarly, the European Systemic Risk Board issued re-ports on systemic risks in the insurance sector (ESRB 2015) and on macroprudential policy beyond banking (ESRB 2016).

9 The amount that an insurer sets aside to fulfill its insurance obligations and settle all commitments to policyholders and other beneficiaries aris-ing over the lifetime of the portfolio, including the expenses of admin-istering the policies, of reinsurance, and of the capital required to cover the remaining risks.

10 These vulnerabilities could be subject to further differentiation regard-ing the various general business models in insurance, such as life insur-ance with minimum guarantees, life insurance without guarantees, non- life short-tail insurance, non- life long- tail insurance, and nonpro-portional reinsurance (IAIS 2013e).

11 This section draws heavily on Jobst 2014.12 Banks assume two major risks—credit risk from lending activities

and liquidity risk from borrowing over the short term and lending long term. These two risks are highly correlated with the economic cycle.

13 This complicates the fair valuation of economic performance (via actu-arial methods) given that insurers receive cash (as gross written pre-mium) for a promise to satisfy an uncertain financial obligation (that is, pay a claim) at an unknown future date.



insurance activities in the transmission of system- wide shocks through interconnectedness and asset liquidation:16

• Business model and cyclicality: Insurance firms are ex-posed to risks commonly found in other financial institutions, including credit, operational, and mar-ket risks (including movements in interest rates and foreign exchange rates), all of which are correlated to varying degrees with changes in economic condi-tions. However, insurance risk (for example, mortal-ity, morbidity, casualty, and liability risks) is largely idiosyncratic and generally independent of the eco-nomic cycle (Box 17.1), which allows them to realize diversification gains (through underwriting inversely correlated risks, such as death insurance and pension insurance, risk pooling, or reinsurance/retrocession). Conversely, banks, by the acceptance of deposits and granting of loans, might find it more difficult to

unexpected cash outflows (due to asset price shocks and/or high surrender rates), and their major source of income (pre-mium inflow) is insulated from market volatility (IAIS 2013a).14 Payouts resulting from claim obligations are nor-mally “managed” in a stress situation, which reduces the speed of cash outflows and allows insurers to hold less liquid assets to support reserves.15

The main differences between banks and insurance com-panies are also apparent with regard to their functional char-acteristics within the financial system, the sensitivity to changes in key macro- financial variables, and their funding structure. These differences explain a more limited role of

14 Institutional failures of insurers have a different impact on the financial system than those in the banking sector, and the way in which they might propagate systemic risk. Insurers tend to have a low level of sys-temic interconnectedness, and their products are not highly complex, which limits systemic risk from a nonsubstitutability of insurance ca-pacity offered by a failing institution (Jobst 2014).

15 Prudent levels of loss reserves—together with the management of the loss adjustment/claims verification processes—help mitigate vulnera-bilities from the risk of sudden outflows of (claims) payments.

16 See Geneva Association 2010a, 2010b, and 2012; and IAIS 2011a and 2012b for a thorough review of the possible systemic relevance of insur-ance activities.

Surplus

Enhanced Capital Requirement(for example, 120 percent of PCR)

Prescribed CapitalRequirement

MinimumCapital

Requirement

Early warning:additional reporting

may be required

Statutoryinsolvency

Solv

ency

Con

trol

Lev

els

Assets

TechnicalProvisions

StatutoryCapital

andSurplus

Enforcementmeasures

Risk Margin (for example, 2 percent of reserves)

Reserves(best estimate)

Source: IMF staff. Note: PCR = prescribed capital requirement.

Figure 17.3 Stylized Insurance Balance Sheet and Solvency Control Levels



of the financial system both domestically and across national boundaries. While the lower degree of in-terconnectedness reduces the negative externalities of failure, it might still pose challenges in their re-solvability, especially for large and complex insur-ance firms.18

• Prudential requirements against risk generation and potential moral hazard: Insurance regulation univer-sally limits insurers to the underwriting of risks that

reduce credit risk (from lending) or liquidity risk (from the maturity mismatch in borrowing short and lending long), especially under stress scenarios.17

• Integration in financial sector infrastructure: As insur-ance firms are not part of payments and clearing sys-tems (which they access but do not have responsibility for organizing), they tend to hold only limited direct intrasystem claims and liabilities and exhibit rela-tively low levels of interconnectedness with the rest

17 Those insurers that failed during the financial crisis did not do so be-cause of their insurance functions, but because of the quasi- banking activities that they engaged in.

18 Even if an insurer does fail, the run- off process takes place over an ex-tended time period, which allows for orderly planning as part of stable processes that do not lead to destabilizing runs.

Box 17.1. General Macro- Financial and Systemic Risk Implications for Insurance1

Although economic cycles impact the investment income and underwriting performance of insurance companies over time, macro- financial linkages vary by business lines as well as technical factors influencing the pricing and reserving of insurance products.

Certain life insurance activities exhibit a high correlation with economic volatility, mainly because of their reliance on stable investment returns to match expected claims over the long term. Higher asset leverage than non- life insurers and longer- duration investments make life insurance companies more susceptible to secular changes in credit spreads and interest rates (unless they are sufficiently hedged). For instance, lower interest rates not only heighten the reinvestment risk for new funds generated from premiums but also increase the pres-ent value of future claims, which could give rise to critical asset- liability mismatches despite temporary asset valuation gains (which might not have a positive impact on profitability after all if investments are held to maturity). Lower investment income might force some life in-surance companies to lower guaranteed returns.2

In the non- life insurance sector, underwriting performance broadly tracks economic growth as available capacity and future pricing ad-justs to changing demand and cost of capital. Large catastrophe losses tend to be followed by premium hardening due to lower insurance capacity; the cost of replenishing underwriting capacity during this “insurance cycle” translates into higher premiums, which is accentu-ated by economic downturns when rising risk aversion of investors and depressed asset prices raise the cost of capital. Conversely, excess capacity would push pricing lower on renewals, which could be accelerated if this cycle coincides with an economic boom with lower cost of capital. Such price dynamics are also influenced by the extent to which renewal rates trail expected underwriting losses. Selective price increases become more likely if the long- term loss trend outpaces historical price increases at the margin.3 In addition, higher rates of infla-tion during periods of economic recovery can adversely affect provisioning and reserve adequacy, especially if changes in claims activity negatively impact performance in real terms.4 However, more than cyclical factors influencing the scale and frequency of different under-writing risks associated with property, casualty, and professional business lines, the erratic occurrence of natural catastrophes and man- made disasters explains significant changes in underwriting performance, whose system- wide impact is driven by firm- specific and/or cross- sectional concentration of exposures.

Some nontraditional forms of life insurance are inherently more susceptible to cyclical effects than are mainstream individual life insur-ance businesses. Capital market- based funding arrangements (such as repurchase agreements, security lending, and over- the- counter derivatives) might require more liquidity over shorter time periods than insurance claims. For instance, the liquidity risk of high- quality collateral to satisfy margin calls for derivatives transactions or cash collateral reinvestment in securities financing transactions differs mark-edly from long- term cash- flow projections associated with insurance liabilities.5 The cash- flow models for security lending and repurchase agreements are generally derived from mark- to- market valuations and can give rise to margin calls if funding liquidity deteriorates. Also, insurance- backed contracts, such as institutional investment and third- party asset management products, such as guaranteed investment contracts, imply some liquidity risk to the extent that policyholders could surrender their contracts at short notice with limited penalties causing a cash- flow scenario comparable to a bank run if contracts are surrendered on short notice.6

1 This box draws on Jobst 2012 and IAIS 2013a.2 During the financial crisis, however, several mitigating factors allowed life insurers to mitigate investment risks. In most cases, the realization of

such adverse effects can be reduced by regulatory forbearance, product designs, and/or personal tax regimes.3 While the growing popularity of insurance- linked securities on natural catastrophes increases the linkage of some insurance firms to capital

markets, the outsourcing of insurance risk via alternative risk transfer mechanisms has arguably muted the impact of the insurance cycle on some business lines, and by extension, has reduced the potential for economic conditions to exacerbate pricing pressures.

4 Thus, an inflationary effect beyond expectations (which implies higher nominal insurance cover due to price appreciation) could cause insurers being under- reserved for future claims.

5 Large transactions of liquidity swaps could make the liquidity position of insurers worse by reducing available cash and liquid assets significantly.

6 For instance, in August 1999, holders of guaranteed investment contracts issued by General American Life Insurance Co. exercised put options that required the life insurer to rapidly repay principal and interest, causing its parent firm, General American Group, to go into administration.



approach to ensure that they can meet their poli-cyholder obligations arising from underwriting risk (especially for non- life insurance firms), which is largely idiosyncratic and generally independent of the economic cycle. Cash inflows from un-earned premiums are invested such that payments of future (unsure) claims can be made at all times, and cash flows so that cash flows from assets and liabilities are matched as closely as possible. Given the scarcity of sufficiently long- term assets, how-ever, insurers tend to have a negative duration gap (“ short- long mismatch”).24

– Cash- flow management: Insurers are generally not predisposed to sudden cash withdrawals, as most insurance liabilities are not redeemable on de-mand by policyholders (like bank deposits). How-ever, excessive lapse risk can arise from adverse economic conditions. For instance, higher interest rates may trigger higher lapse rates as more policy-holders switch to other products for higher return, which may result in potential loss caused by sell-ing investment assets for cash (or other assets) needed to cover surrender payments.

However, the long- term funding structure of insurers puts a premium on the valuation methods for best/current esti-mates of liabilities. Actuarial pricing and loss reserve models underpinning the valuation of liabilities might not fully re-flect the stochastic properties of risk factors, especially during times of stress. In addition, some models require sufficient empirical observations over long periods of time. Examples include calibration errors in estimating mortality and lapse rates in life insurance, catastrophe risk, and insurance poli-cies on perils with few empirical observations, such as pan-demics or terrorism, or perils that materialize only over long periods of time, such as asbestos- related claims.

3. PROCESS AND METHODOLOGIESAn effective design and implementation of the stress testing process entails a series of considerations that inform the scope of analysis, the choice of the methodology, and the interpretation of relevant findings for potential supervisory follow- up and capital planning (Box 17.2). The specific steps include: (1) determining the object of analysis (structural conditions, regulatory situation); (2) defining the scope; (3) developing the methodological framework and data analysis; (4) considering the valuation standard and the treatment of capital resources; (5) designing stress scenarios; (6) selecting the appropriate risk factors; (7) defining the output measures;

represent insurable interest.19 Therefore, insurers cannot generate additional risks but rather aim at managing and controlling existing risks over a de-fined time horizon.20 They tend to retain risks on their balance sheets (as opposed to banks’ transfer of risk), mitigating the risk of moral hazard. The rein-surance of primary underwriters and the acceptance of ceded insurance risk between reinsurers involve only a partial risk transfer (that is, most risk remains on the ceding [re]insurer’s balance sheet). In addi-tion, while the trading of derivatives (for example, credit default swaps) could be made in the absence of insurable interest, reinsurance generally transfers clearly defined risk (by way of indemnification) and is inherently linked to the insurable interest ceded by the policyholder.21

• Funding structure: Strong operating cash flows due to upfront premium payments (“inverted production cycle”) shield insurers’ liquidity positions from exter-nal funding conditions. This prepaid funding model (with the possibility of continued collection of pre-miums even in a recovery or resolution phase), to-gether with the longer duration of the claims process and penalties for early surrenders of (life) insurance policies (“lapse risk”), makes insurers less susceptible to liquidity runs.22 Claims can normally be paid via the sale of liquid assets that generate commensurate cash inflows (as opposed to traditional financial in-termediation, which involves maturity transforma-tion).23 Thus, insurers can become insolvent (or insufficiently solvent) and still remain liquid due to the long- term nature of the business model (which lacks business activities that are more subject to mar-ket risk, such as maturity transformation, consumer or commercial credit, and transaction clearing services). However, liquidity risk can arise from asset- liability mismatches and cash- flow management:– Asset- liability matching: Insurance companies pur-

sue a predominantly liability- driven investment

19 The IAIS (IAIS 2011a) defines insurable interest as “an interest in a per-son or a good that will support the issuance of an insurance policy; an interest in the survival of the insured or in the preservation of the good that is insured. […] Financial derivatives are not considered insurance for regulatory purposes.”

20 Reinsurance shares certain characteristics with derivatives transactions, with the latter being generally classified as a noninsurance activity (IAIS 2012b).

21 Note, however, that insurance contracts with limited or no risk transfer can change the risk profile, making at least part of the insurance trans-action nontraditional or even noninsurance (IAIS 2011a).

22 Most life insurers also secure longer term and well- diversified retail funding compared with other types of financial institutions.

23 Even though some forms of life insurance may be viewed as savings products, most contracts have tax and contractual disincentives for policyholders to surrender the insurance policies before its contractual maturity (that is, insurance reserves are not instantaneously “puttable” like deposits). Conversely, where reserves are “puttable,” the policy-holder bears the investment risk ( unit- linked, separate accounts).

24 This implies that insurers generally benefit from rising interest rates (es-pecially long- term business) whereas banks tend to experience valuation losses on investments, which might outweigh the benefits from higher interest margins as interest rates rise and funding costs adjust only slowly.



(8) validating the results; and (9) dealing with the outcome of the stress test (Figure 17.4).

This section focuses on the conceptual discussion of these core elements, which is complemented with a review of rele-vant characteristics of national supervisory practices in in-surance stress testing and the experience of the IMF staff in FSAPs. While publicly available information is limited, the analysis includes 15 jurisdictions and regional bodies that have disclosed some specifications for their stress testing ex-ercises. The IMF experience includes 27 FSAPs, which have been completed since 1999 in both advanced and developing economies (until the end of 2016).

Object of Analysis

A thorough insurance market analysis of external factors and general business conditions impacting firm behavior should be conducted, which includes acknowledging signifi-cant differences in business models, the role of insurers in the domestic financial sector, and international linkages, es-pecially when offshore companies are relevant to the finan-cial system. In the context of FSAPs, this is often covered by the assessment of the ICPs. The following characteristics of

Box 17.2. The Taxonomy of Stress Testing Approaches

Stress testing can serve several different but usually interrelated purposes, which include macroprudential surveillance, microprudential supervision, crisis management, and risk management (see Chapter 15, Figure 15.1). The first three types of stress tests are completed (or requested from insurance companies) by regulatory authorities and can be performed by means of a bottom- up or top- down approach, or as a combination of both. Stress tests with a bottom-up approach are run by companies mainly for internal purposes, but also by credit- rating agencies in the process of assigning or monitoring ratings. They are based on a prescribed set of assumptions and scenarios and providing the results to the supervisory authority, while top- down tests are run by the supervisory authority based on input data provided by companies, for example, via regular reporting channels or via public disclosure.

Macroprudential stress tests determine the system- wide resilience to shocks within the financial sector. They can be limited to one financial sector or provide a cross- sectoral perspective by capturing interlinkages between banks, insurers, and other market participants. System- wide stress tests completed in the context of IMF Financial Sector Assessment Programs generally fit this description.

Microprudential stress tests are used by supervisory authorities to determine firm- specific vulnerabilities to stress. Especially in jurisdic-tions that have not yet introduced a risk- based solvency system and accompanying supervisory reporting, such microprudential stress tests are an important tool that is commonly used by supervisors around the world; some examples are presented in Appendix Table 17.1.4.

Crisis- management stress tests help identify actual or potential capital need of distressed companies. The Supervisory Capital Assessment Program and the first Comprehensive Capital Analysis and Review run by the US authorities in early 2009 as well as the Capital Assessment carried out by the European Banking Authority in 2011 are prominent examples of this type of stress test.

Risk- management stress tests are used by individual financial institutions to manage and plan their business activities in a more forward- looking manner. As the main objective of the tests is for internal purposes, there are a variety of approaches.

The concept of reverse stress tests can also be effective but limits the scope of application. Reverse stress tests set a certain target threshold for an output measure, for example, a solvency ratio of 100 percent, with the aim of identifying the required magnitude of shock to one specific risk factor that would lead to a breach of the threshold criterion. This condition could be relaxed by providing shocks also for other risk factors.1

1 Note that reverse stress tests should be used only for material and relevant risks lest they generate meaningless results when the stressed risk ex-posure is low in absolute terms. For instance, given the low level of equity exposures of many insurance companies, only an implausibly large decline in equity prices would result in a breach of a predefined solvency threshold. Conversely, this method is quite relevant for assessing the impact of large parallel upward or downward shifts of the interest rate term structure, which affects a wide range of interest- rate- sensitive exposures.

Object ofAnalysis

Scope

Methodology

Valuation& Capital

ScenarioDesign

RiskFactors &

Aggregation

OutputMeasures

Validationof Results

Communicatingthe Results

Source: Authors.

Figure 17.4 Stress Testing Process



• Tax regime: Shock absorption effect of deferred tax assets/liabilities; relative tax advantages of savings products offered by insurance undertakings (com-pared to those offered by banks).

• Policyholder protection funds: Existence of a protec-tion fund and its coverage, also in comparison with deposit insurance scheme (if applicable).

Determination of Scope

A capital assessment under stress should capture all material risks affecting the solvency of all relevant legal entities and/or insurance groups. Nonfinancial activities that are nonma-terial could be excluded, but noninsurance financial activi-ties would need to be considered in any case.26 Group participation usually involves foreign businesses, which should be included to examine vulnerabilities to the group and its parent company. However, supervisory authorities usually have less granular information at the group level (with regulatory reporting being focused on firm- level re-sults), which makes TD stress tests more difficult and com-plicates the validation of BU results. If group- wide stress tests are run by the home supervisor, the results should be communicated to and discussed with host supervisors.

The coverage generally depends on the market structure in each specific case and the extent to which NTNI activi-ties are relevant for the specification of spillover and conta-gion risks. Market coverage should be calculated separately for the life and the non- life sector. If the market is highly concentrated, it is usually sufficient to include just the larg-est companies in terms of premiums or assets. However, medium- sized (or even small) insurers should be included if they conduct significant NTNI activities and/or these firms (1) are expected to be highly vulnerable to certain shocks to risk factors; (2) have high relevance for the real economy or the financial sector by offering specific prod-ucts (for example, credit or mortgage insurance); or (3) are very interconnected (for example, as a reinsurer or in a fi-nancial conglomerate).

Most FSAP insurance stress tests have been completed for samples comprising the largest firms only.27 For most ex-ercises, the system- wide coverage ranged between 55 and 85 percent of the market (and averaged about 70 percent) (Appendix Table 17.1.3). Depending on the size and the in-dustry structure within a particular jurisdiction, the cover-age sometimes involved a larger number of firms (for example, 30 each in the case of the FSAPs for Switzerland and the United States 2010 and more than 70 in similar ex-ercises for France and Germany). Thus far, the market cover-age of the insurance sector has been less than that of the

the local insurance sector are relevant for the design and implementation of an effective stress test:

• Insurance business: Prevalence of specific lines of business and profitability of insurance business (breakdown of profit sources); share of traditional versus investment- linked life insurance; existence of country- specific insurance products, especially when these are subject to specific regulations (for example, state- sponsored catastrophe insurance or retirement schemes); reserve adequacy (duration of liabilities, average guaranteed interest rates, and the degree of penalty of early termination of policies); and the use of reinsurance and use of risk- mitigating features like profit sharing.

• Investment portfolio: Asset composition (equity, bonds, loans, real estate, alternative investments [such as private equity], hedge funds, and commodities) with break-down of countries, sectors, duration, and liquidity (Level 1, 2, and 3); and hedging transactions, especially for interest rate risks.

• Connectedness within insurance groups and conglomer-ates: Corporate structure of groups and/or conglom-erates, including foreign group entities and special purpose vehicles; intragroup transactions, for exam-ple, committed funding arrangements (securities holdings/lending with parents or other group com-panies) and intragroup reinsurance activities; and possible interventions by local supervisors, such as ring- fencing.25

• Interlinkages to other financial institutions: Exposures to other entities within the financial sector, typically through asset exposures (equity, senior bonds, sub - ordinated debt, and commercial paper), deposits, derivatives transactions, and securities financing trans -actions (securities lending/repos). Usually, NTNI- related exposures (such as derivatives trading and securities financing transactions) are negligible ex-cept for large firms and monoline insurers. Some in-surers (with large investment portfolios) could provide large amounts of liquidity to banks via asset swaps (and other funding commitments). Such arrangements are prone to contagion effects during times of stress.

• Solvency regime and prudential standards: Existence of a risk- based regulatory framework (including level of confidence, time horizon, and calibration of risk factors) as well as risk coverage.

25 Intragroup transactions within conglomerate structures often involve more liquidity risk under stress. Against a backdrop of an overall loss in confidence in capital markets, the banking side of conglomerates (or bank counterparties to liquidity swaps) could become vulnerable to the risk of large withdrawals of deposits and/or the runoff of liabilities. As both banks and insurers would sustain a sharp decrease in the value of their investment portfolios, funding needs could lead to greater reli-ance on intragroup transactions (or the use of contingent funding arrangements).

26 In the same vein, off- balance sheet exposures need to be considered (consistent with a total-balance-sheet approach).

27 Exceptions were Luxembourg and Israel, where all domestic insurance companies were within the scope of the stress test.



On investments, few supervisory authorities have the de-tailed information needed to conduct TD stress tests, such as information on the duration and credit quality of bonds or detailed information on hedging transactions. On the li-ability side, information on reinsurance programs and the duration of liabilities is often missing. In the absence of such critical information, a BU stress test would usually be a nec-essary first step to gather information on the range of values for the key parameters used in a subsequent TD stress test. A robust modeling framework for a TD stress test would also require a global valuation and solvency regime together with harmonized reporting and disclosure standards, which are still in development. Therefore, insurance supervisors have become increasingly reliant on a mixture of both TD and BU approaches to stress testing.30 FSAPs have predomi-nantly applied BU stress tests, with just five exceptions (Ta-ble 17.1 and Appendix Table 17.1.3).31 While prudential data are commonly used for FSAP insurance stress tests, this ap-proach does not preclude the use of public data for the prep-aration of BU exercises and the cross- validation of stress test results, such as in the case of Japan (2003) and the United States (2015).32

The data used in the stress test should be sufficiently gran-ular to account for vulnerabilities from intragroup transac-tions or transactions between banking and insurance legal entities. Most supervisory stress tests require firms to report results on a legal- entity basis (especially when the solvency regime supports group- wide oversight, such as in the case of Bermuda, Germany, Switzerland and United Kingdom). In many cases, however, stress tests are completed only on a con-solidated reporting basis (European Union), which does not cover the assessment of the impact of intragroup transactions on the capital and liquidity positions of legal entities that are part of a group or conglomerate. If intragroup transactions and transactions between banking and insurance entities are salient risk drivers, these sources of vulnerability require more granular prudential information, which would need to be explored on a legal- entity basis (together with a greater supervisory involvement). Thus, some stress tests in FSAPs are completed on both a solo and a consolidated reporting basis.

banking sector in FSAP stress tests (Jobst, Ong, and Schmie-der 2013), which can be explained by the fact that smaller insurers have only a marginal impact on financial stability unless they are highly interconnected through reinsurance or NTNI business.

Methodological Framework and Data Quality

The choice of a suitable stress testing model(s) and technique(s) needs to be proportionate to the nature, scale, and complex-ity of the insurance sector.28 Insurance stress tests are tradi-tionally completed as BU exercises, which is reflected in the growing number of guidance and consultation papers by supervisors, industry, and international organizations. The IAIS (2003) was first to propose a standardization of the design and implementation of supervisory stress tests to es-tablish greater consistency of insurance risk scenarios, and in 2013 provided some additional guidance in its paper on MPS in insurance (IAIS 2013a). Shortly thereafter, the In-ternational Actuarial Association (IAA 2013) published a paper that provided an actuarial perspective on scenario analysis and stress testing.29 There is also noticeable prog-ress at the national level. In Europe, insurance stress test-ing has been advanced by the European Insurance and Occupational Pensions Authority (EIOPA 2011a, 2011b, 2014, 2016); the EU insurance regulator conducts regular BU stress tests with customized scenarios, which build on the first system- wide supervisory stress test for the Euro-pean insurance sector (CEIOPS 2009, 2010). In the United Kingdom, the Prudential Regulatory Authority (PRA 2013) issued a statement on its approach to insurance su-pervision, which also included several references to stress testing, and the implications for supervisory assessments of capital planning. This followed an earlier consultation pro-cess with industry (FSA 2008) on scenario analysis and stress testing, which concluded in a statement on the use of stress tests within the prudential regulatory regime for insurers (FSA 2009).

Data constraints are the biggest challenge for TD insur-ance sector stress tests, which are generally less developed than BU approaches. While the main advantage of TD exer-cises lies in the swift availability of results, data limitations often hinder the development of these approaches in regular supervision. Usually, data input for stress testing exercises stems from regular prudential reporting, which can differ widely across countries in terms of scope and granularity.

28 More sophisticated models increase the chances of estimation uncer-tainty, which needs to be considered when drawing policy conclusions from stress tests.

29 The paper presented survey results on current stress testing practices to develop recommendations on how stress tests could be applied to de-fined contributions plans. The assessment of the economic impact of risk factors on defined contribution plans is similar to that of life insur-ance companies (Ionescu and Yermo 2014).

30 In 2013, the European Central Bank introduced a market- consistent framework for monitoring the stability of large euro area insurance groups (ECB 2013; Vouldis and others 2013). A macroeconomic sce-nario affects insurance companies via valuation changes for both assets and liabilities, potential sale of assets due to a cash- flow drain caused by higher lapse rates, and changes in the credit quality of the loan portfo-lio. The result of the stress test can be presented in terms of total balance sheet assets as well as net assets, with the latter serving as a proxy for a solvency measure.

31 For Israel and Germany, only a TD exercise was performed, and in Portugal and South Africa (2008), a TD exercise complemented the BU approach.

32 Data quality also imposes practical limitations to comparative analysis. High data granularity, while giving deeper insight into the mechanics of results, can also increase the potential for different interpretations, straining the analytical poignancy.


Andreas A

. Jobst, Nobuyasu Sugim

oto, and Timo Broszeit

455

TABLE 17.1

Stress Testing Matrix (STeM): Stylized Summary of Insurance Stress Testing Approaches in IMF FSAPs and National Supervisory Frameworks1

Components Key Elements Explanation IMF FSAPs National Supervisory Approaches

1. Scope Approach Top-down (TD)/bottom-up (BU) Completion of exercise by supervisor/FSAP

team (TD), aggregation of individual results received directly from firms (BU)

Mostly completed as BU exercise (with the occasional use of TD to complement BU results and as basis for sensitivity analysis)

Coverage and relevance

Institutions, market share Number/type of insurance companies, percentage of insurance sector assets or

premiums

Usually largest firms (four–six firms in smaller countries, but more firms in larger countries [for example, 30 firms in the case of the United States]); usually both life and

non life firms (but depends on signifi-cance), between 43 percent and 100

percent

Usually all firms, but coverage varies between 50 percent and 100 percent

Data Source Insurer’s own, prudential, and/or public data Mostly prudential data (but also public data, especially for TD)

Mostly insurer’s own data (but also prudential data, especially for TD)

Scope/reporting basis

Reference basis for economic impact (solo vs. consolidated),

cut-off date

(Un)consolidated insurance group or domestic business only

Mostly solo basis, but also more consolidated reporting recently; end

of fiscal year

Mostly solo basis, but also consolidated reporting; end of fiscal year

2. Valuation Basis Assets/liabilities Market-consistent or statutory

accountingDefines the degree of market- consistency (fair value assets and best estimates of liabilities/

technical provisions)

Mostly statutory accounting, but recently more instances of market-consistent

valuation

Statutory accounting

Confidence level Measure of statistical accuracy VaR, CTE Rarely specified VaR (up to 99 percent),CTE (up to 99 percent)

3. Scenario Design2

Macro-financial linkage/transmission channel(s)

Single-factor shocks; macro-financial conditions influencing investment and underwriting performance

Shocks are defined based on the aggregate impact of individual stresses on identified risk factors, joint impact from one or more adverse economic scenarios (as the result of changes in certain equilibrium conditions); also sensitivity analysis of certain risk drivers help assess the

robustness of estimates to changes in the severity and combination of risk factors

Combination of single-factor shocks mostly without specification of general macroeco-

nomic conditions/scenarios

Combination of single-factor shocks without specification of general

macroeconomic conditions/scenarios; in some instances contains macro-financial

linkages of capital market shocks

Risk horizon Single period (instantaneous, or multiple-period forecast

horizon)

Forecast horizon over which the severity of stresses are applied (also determined by

maturity term of liabilities)

Single-period (one year) stress horizon, but most recently multiple periods

(two–three years)

Mostly single-period stress horizon with shocks prescribed by supervisor

(continued)



456




4. Risk Factors3

Assets Freq.(H/M/L)

Freq.(H/M/L)

Credit/ counterparty risk

Market value changes of fixed income instruments, increase of counterparty risk, and economic value change of loan portofolio

Relative/absolute increase of sovereign credit spreads (at different maturities, rating grades)

based on benchmark corporate/sovereign debt and/or credit default swaps for specific maturity tenors at a given level of statistical confidence,

possibly combined with the assumption of higher implied volatility

Increase of credit spreads by up to 50 percent, downgrade of counterparties by two–four

notches, impairments consistent with the realization of implied

PD of Basel II risk-weights, failure of large counterparty

H Rating-class specific increase in credit spreads (but often

unspecified)

H

Equity risk Market value changes of equity and alternative investments

Uniform decline in market values About -30 percent (but up to -50 percent)

H About -20 percent (but up to -40 percent)

H

Assets Freq.(H/M/L)

Freq.(H/M/L)

FX risk Negative/positive shocks to net open FX positions and/or

FX-denominated assets and liabilities

Significant FX rate appreciation/depreciation (for example, multiple of historical volatility of

FX rate pairs under stress)

Around +/-20 percent (but up to +/-50 percent)

H Around +/-20 percent (but often unspecified)

H

Real estate risk Economic value change of exposures sensitive to real estate

values

Uniform drop of real estate prices About -20 percent (but up to -50 percent)

H About -15 percent (but often unspecified)

H

Interest rate risk Economic value change of interest-sensitive assets and

liabilities

Change/shift of risk-free yield curves of domestic and foreign currencies (parallel, steepening, flattening)

About +/-200 bps parallel shift H About +/-100 bps parallel shift (but often unspecified)

H

Liabilities Freq.(H/M/L)

Freq.(H/M/L)

Life underwriting Mortality/

morbidity/longevity

Economic loss caused by higher mortality, morbidity, and longer

life expectancy; catastrophe- related risks from pandemics

included in this category

Revaluation of technical provisions (reserves) due to longer claim periods and/or higher

claim frequency

Mortality/morbidity/longevity of annuitants (about +25 percent each); occasional

testing of pandemic

M Included in most countries with significant life insurance

business (but severity not disclosed)

M

Lapse risk Economic loss caused by higher surrender rates

Share of policies surrendered prematurely; share of policies that result in underwriting

losses due to higher lapse rates

Mass lapse of about 25 percent (but up to 50 percent)

L Rarely included (severity not disclosed)

L

Non-life underwriting

Natural catastrophe

Economic losses from natural and man-made disasters

Perils related to windstorms, earthquakes, wildfires, floods, and terrorism

Usually set to historical catastrophic events or

maximum historical claims experience, such as 1-in-100

years probable maximum loss, or aggregate policy limit

M Usually defined as peak risk based on internal (firm-

specific) models or industry benchmarks (often at very

high levels of statistical confidence, such as 1-in-200

years)

M

(continued)


Andreas A



457



Components Key Elements Explanation IMF FSAPs National Supervisory Approaches Other non-life

underwriting shocks

Cost/claim increase Relative impact/severity of premium/reserve risk, misestimation of liabilities (especially

frequency and cost of claims)

Large variation in assumptions, but usually around +10 percent average cost of claims and +15

percent higher frequency of claims

L Mostly focused on premium risk and frequency of claims (but no severity disclosed)

M

Other risk factors Freq.(H/M/L)

Freq.(H/M/L)

Deterioration of perceived risk profile

Rating downgrade Relative impact/severity of collateral require-ments, loss payment triggers on in-force policy

contracts, claw-backs, and/or other adverse financial and liquidity implications of the

downgrade

n/a — Rating downgrade of insurer by [x] number of notches, off-balance sheet items

L

Second-order effects

Feedback effects; management and regulatory action

Consideration of feedback effects that compound the impact of risk factors as well as

operational/strategic change(s) to business model due to shock

n/a — Mostly focused on manage-rial actions and capital

planning

M

Other risk factors Freq.(H/M/L)

Freq.(H/M/L)

Combination of financial/underwrit-ing scenarios

Lower premiums after instant shocks; coincidence of peak

underwriting losses and asset price depreciation

Lower premium after policyholder reduction of dividends; combined insurance and capital market shock proxies liquidity risk (especially,

short-tail business)

Only as part of ex post sensitivity analysis (but rarely

used thus far)

L Mostly firm-specific scenarios

L

Risk mitigation (reinsurance and hedging)

Reinsurance and derivatives Interest rate swaps for mismatches, reinsur-ance/retrocession agreements, alternative risk transfer (insurance-linked securities, side cars,

embedded value securitization)

n/a — Mostly firm-specific scenarios, with/without

hedging assumption

H

Risk aggregation/diversification effects

Diversification among risk factors and entities

Correlation assumptions among various risk factors (for example, diversification benefit in

Solvency II standard formula)

Various simple summations of individual single-factor shocks;

sometimes use of correlation matrices for diversification

effect

M Various simple summations of individual single-factor

shocks but rarely application of diversification effects

M

5. Regulatory Capital Standards Capital definition/

solvency requirementMetrics (regulatory solvency/premium/loss reserve ratio)

Minimum solvency margin requirement/minimum capital requirement/prescribed

capital requirement/enhanced capital requirement, or other general accounting-

based/risk-based solvency standard including premium/loss reserve ratio and excess assets

over liabilities (net asset value), with standard-ized charges (solvency capital requirements) or

charges based on approved internal model results

Application of existing prudential solvency standard, but also alternative stress testing

measures, such as loss measured as percentage of shareholder equity, minimum

regulatory premium/loss reserve ratio, solvency margin ratio, net asset value

Existing prudential solvency standard (for example, Solvency I, Minimum Continu-ing Capital and Surplus Requirements,

Bermuda Solvency Standard, Risk-based Capital, Swiss Solvency Standard,

Singapore Risk-based Capital, Individual Capital Adequacy Standards, Solvency

Margin Ratio)

(continued)



458




Capital adequacy Threshold Capital and surplus/amount of recapitalization (in domestic currency) based on choice of

solvency requirement/ ”pass mark” for stress test

For example, 100 percent solvency level after application of mitigating factors (if applicable), such as diversfication effects; ECR: 120 percent of MCR or MSM: higher

of USD $[x] or [x] percent of net written premiums, and/or [x] percent of technical provisions

6. Methodology Stress test model Accounting-based or market-

based (economic)Determines the degree of market- consistency of economic loss estimates and implications

for solvency assessment

Balance sheet approaches with varying degrees of conservativeness/actuarial

assumptions; systemic contingent claims analysis (Jobst and Gray 2013) as market-

based technique

Dominance of actuarial approaches based on supervisory guidelines; rising acceptance of economic balance sheet

approaches

Modeling of risk factors Asset/insurance risks; macro- financial linkages

Calibration/parameterization of risk factors affecting both assets and liabilities under stress using market information, historical

experience, and expert judgment

Adaptation of existing supervisory approaches with sensitivity analysis

regarding specific parameters

Reliance on firm’s vendor models (especially for non-life business) and

internal approaches; econometric models for income elements and lapse rates

7. Communication Presentation of output Template(s) Standardized output template for individual

resultsSee Figures 17.7a and 17.7b

Publication Internal (with authorities), external

Results published in FSSA (and Technical Note) See Appendix Tables 17.1.1 and 17.1.2

Source: Jobst, Sugimoto, and Broszeit 2014.Note: The template for these output charts is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit. ALM = asset-liability management; bps = basis points; CTE = conditional tail expectation; ECR = enhanced capital requirement; FSSA = Financial Sector Assessment Program; FX = foreign exchange; MCR = minimum capital requirement; MSM = minimum solvency margin requirement; PCR = prescribed capital requirement; PD = probability of default; VaR = value at risk.1This table was originally presented in IMF Working Paper 14/133 (Jobst, Sugimoto, and Broszeit 2014). 2The scenario design also includes factors that management of insurers can control, such as balance sheet growth, dividend pay out, and other business strategy considerations. In FSAPs, common assumptions are that the balance sheet growth is in line with nominal GDP, the firm maintains its historical dividend pay out ratio over the forecast horizon, and there are no changes in investment portfolio, funding sources, and business model/underwriting behavior. National approaches benefit from greater insight on the supervisory implications of managerial actions, but assumptions in most approaches are consistent with those applied in FSAPs.3“Freq. (H/M/L)” denotes the frequency of each risk factor in IMF and national stress tests, respectively, where H = high (always/nearly always), M = medium (frequent), and L = low (rare/never).


https://www.elibrary.imf.org/page/stress-test2-toolkit. ALM = asset-liability management; bps = basis points; CTE = conditional ta




ate a more market- consistent valuation of both assets and liabilities. This is because factor models (with predetermined and constant parameters) may not be able to fully capture time- varying and/or nonlinear risks. Figure 17.5b shows that a similar comparison can be made for the valuation aspect of a solvency regime considering the impact of assumptions af-fecting the level of transparency, including initial conserva-tive and ex- post adjustments. Jurisdictions that have moved to market- consistent valuation accept greater procyclicality to better align capital adequacy with the economic cost of capital (and the way it influences management decisions, and, if need be, supervisory enforcement activities).

There are three major valuation approaches that are found in existing solvency regimes:

• Accounting basis (for example, Solvency I in the EU): The valuation is based on historical prices (that is, cost accounting) without consideration of the actual risk. The absence of risk- based elements affecting the valuation makes this standard less suitable for the quantification of the economic impact of changes in asset prices and interest rates.

• Risk- based approaches (for example, Risk- Based Capital in the United States, Solvency II in the EU): For Sol-vency II, the valuation basis includes alleviations, such as the dampeners of the long- term guarantee package (that is, volatility adjustment, matching adjustment, and the convergence period for extrapolating the basic risk- free curve), which decrease the sensitivity to inter-est rate and spread changes, resulting in lower techni-cal provisions and higher own funds.

• Market- consistent valuation basis (for example, Swiss Solvency Test): For technical provisions, cash flows due to insurance liabilities are discounted with an appropriate risk- free rate based on asset swap rates (after controlling for credit risk) or replicated with sovereign bonds only (without including spread risk premia as in many risk-based approaches).35 Assets are valued based on available market prices.

The market- consistent (rather than cost- based) valuation approach offers the most objective and economic view of assets and liabilities. It generates a transparent fairvalue rep-resentation of assets based on reliable market prices and best estimates of insurance liabilities. Robust validation is neces-sary to minimize model risks and valuation uncertainty, which would increase capital or lower technical provisions under the disguise of greater robustness.

Valuation and Capital Resources

The capital assessment under stress reflects the interaction be-tween the macro- financial impact of risk factors and the characteristics of the relevant solvency regime in a country. The results of the stress test project the balance sheet impact for different adverse scenarios based on the risk sensitivity and the valuation of exposures. All assets and liabilities of insurers would need to be appropriately and consistently val-ued. Typically, valuation standards prescribe the fair valua-tion of investment assets and the best actuarial estimate of insurance liabilities (including adequate reserves and pro-visions). Cash- flow projections should incorporate future demographic trends as well as legal, medical, technological, social, and economic developments, with appropriate as-sumptions relative to the relevant exposure, gross of reinsur-ance and special purpose vehicles. Discount rates applied to cash flows should be consistent with observable market prices for investments whose cash flows match those of insurance liabilities in terms of timing, currency, and liquidity.33

The risk sensitivity of the solvency regime and applicable valuation standards influence the selection of inputs, the esti-mation of changes in liability- matched asset values and techni-cal provisions, and the interpretation of the stress test results.34 Some solvency regimes contain alleviations based on conserva-tive assumptions of asset prices and best- estimate liabilities (in-cluding discount rates). These alleviations trade off greater robustness of the capital assessment (by reducing the procycli-cality of valuation) against the risk of understating insurance liabilities derived from less reliable market prices (which might include liquidity and/or credit risk premiums). These differ-ences vary across countries and can significantly impact capital resources, risk measurement, and solvency positions of insurers.

A comprehensive analysis of the risk measurements of capital assessments and applicable valuation standards should be made prior to developing a specific stress test. Figures 17.5a and 17.5b visualize the most important features of sol-vency regimes in major (re)insurance markets— Bermuda, the EU, Japan, Switzerland, and the United States. Fig-ure 17.5a compares the degree of market consistency and the level of economic realism implicit in risk measurement under different solvency regimes. The prevalence of internal model approaches— if combined with a rising degree of statistical confidence of risk measures, few (or no) diversification bene-fits, and a comprehensive scope of reporting— tends to gener-

33 They should exclude the effect of the insurer’s nonperformance risk to avoid introducing noneconomic volatility in net assets.

34 The consistent use of valuation standards also facilitates comparability of the solvency impact on balance sheets of internationally active insur-ance groups operating under different solvency regimes in the absence of global accounting and actuarial standards supporting a consistent capi-tal requirement. Currently, the comparability of valuation standards and capital treatment across countries is complicated by the divergence of ex-isting accounting standards, the capital treatment of off- balance- sheet items, and the intended scope of covering all financial activities within insurance groups (which could raise consistency issues vis- à- vis other capital regimes).

35 For instance, the following specifications are commonly used for the bal-ance sheet valuation and the calculation of solvency capital requirements. The term structure of discount rates is extrapolated based on the Smith- Wilson method (or similar), assuming a certain ultimate forward rate, with a downward adjustment to the calculated forward rate to account for credit risk, and is then applied to cash outflows related to future insurance liabilities over the same time horizon (EIOPA 2018). EIOPA published a tool for European insurers to derive the UFR at https://eiopa.europa.eu /Publications/Standards/20180813_Technical%20Documentation%20%28RP%20methodology%20update%29.pdf.


https://eiopa.europa.eu/Publications/Standards/20180813_Technical%20Documentation%20%28RP%20methodology%20update%29.pdf




in insurers’ balance sheets, the application of cost accounting and, to a lesser extent, other adjustments that assume hold- to- maturity investment to derive the discount factor for liabili-ties would be inconsistent.37

Alleviations to market- consistent valuation (and simplifying assumptions based on cost accounting) would understate the economic impact of market stress scenarios under a risk- sensitive

Most insurance stress tests completed by national supervi-sors allow only (or require) the use of historical cost account-ing for insurance liabilities (“ cost- based valuation standard”) while assets are usually valued in a market- consistent man-ner.36 The statutory valuation at historical cost is premised on the assumption that insurers can continue holding most as-sets until their maturity to generate sufficient cash flows to pay claims and other liabilities over the same time. However, given the potential and actual turnover of investment assets

36 Market- consistent valuation of insurance liabilities is used only in a few cases (for example, Switzerland).

37 The effect is even more pronounced for stress test results under Sol-vency I. Solvency I ratios contain simplifying/conservative assumptions that react very weakly to stresses, leading to generally very stable Sol-vency I margins.

Leve

l of E

cono

mic

Con

sist

ency

/Rea

lism

Figure 17.5b. Overview of Solvency Regimes: Valuation Standards

Full internal model approachwith stringent calibration requirements

and down-cycle emphasis

Accounting/balance sheet approacheswithout consideration of variables risks

and volatility of exposures

Risk-based approacheswith economic balance sheetassumptions and adjustments

Japan SMR• Solo, branch, consolidated

• Factor-based at 95.0 percent VaR (assets) and >99.0 percent VaR (liabilities)

Solvency I• Consolidated

• Balance sheet/leverage ratiowithout consideration of actual risk

Solvency II• Consolidated

• Standardized/internal model at 99.5 percent VaR• Diversification benefit

BSCR• Consolidated

• Factor-based/internal model at99.0 percent CTE

• Diversification benefit

SST• Solo1

• Internal model at 99.0 percent CTE

RBC• Solo basis

• Factor-based at 98.0 percent VaR

Partial internal model approachwith limited reliance on firm-based estimates

Market consistency/Risk sensitivity

Leve

l of T

rans

pare

ncy Market-consistent

with best estimate liabilities and fairvalue assets, without alleviations

Lock-inWith simplifying/conservative assumption

RBCLock-in with CF testing (ICP 14)

• Conservative assumptions• Full lifetime analysis in ex-post

revaluation

Solvency I• Predefined capital requirements

• Balance sheet valued at historical cost

SSTMarket-consistent with

certain alleviations, suchas temporary change of

discount rate

BSCRSome initial assumptions with

certain alleviations

Solvency IIMarket-consistent with certain alleviations, such aspremium matching and counter-cyclical measure

Japan SMRLock-in with CF testing (ICP 14)

• Some initial assumptions• 10-year analysis in ex-post

revaluation


Figure 17.5a Overview of Solvency Regimes: Risk Measurement

Leve

l of E

cono

mic

Con

sist

ency

/Rea

lism

Figure 17.5b. Overview of Solvency Regimes: Valuation Standards

Full internal model approachwith stringent calibration requirements

and down-cycle emphasis

Accounting/balance sheet approacheswithout consideration of variables risks

and volatility of exposures

Risk-based approacheswith economic balance sheetassumptions and adjustments

Japan SMR• Solo, branch, consolidated

• Factor-based at 95.0 percent VaR (assets) and >99.0 percent VaR (liabilities)

Solvency I• Consolidated

• Balance sheet/leverage ratiowithout consideration of actual risk

Solvency II• Consolidated

• Standardized/internal model at 99.5 percent VaR• Diversification benefit

BSCR• Consolidated

• Factor-based/internal model at99.0 percent CTE

• Diversification benefit

SST• Solo1

• Internal model at 99.0 percent CTE

RBC• Solo basis

• Factor-based at 98.0 percent VaR

Partial internal model approachwith limited reliance on firm-based estimates


Leve

l of T

rans

pare

ncy Market-consistent

with best estimate liabilities and fairvalue assets, without alleviations

Lock-inWith simplifying/conservative assumption

RBCLock-in with CF testing (ICP 14)

• Conservative assumptions• Full lifetime analysis in ex-post

revaluation

Solvency I• Predefined capital requirements

• Balance sheet valued at historical cost

SSTMarket-consistent with

certain alleviations, suchas temporary change of

discount rate

BSCRSome initial assumptions with

certain alleviations

Solvency IIMarket-consistent with certain alleviations, such aspremium matching and counter-cyclical measure

Japan SMRLock-in with CF testing (ICP 14)

• Some initial assumptions• 10-year analysis in ex-post

revaluation


Source: IMF staff.Note: BSCR = Bermuda Solvency Capital Requirement; CF = cash flow; CTE = conditional tail expectation; ICP = insurance core principle; RBC = risk-based capital; SMR = solvency margin ratio; SST = Swiss Solvency Test; VaR = value at risk. This figure provides a general comparison of risk measurement and valuation standards (assets and/or liabilities), which informs a stylized scaling of relative economic consistency (Figure 17.5a) and transparency (Figure 17.5b) relative to the risk sensitivity of different solvency regimes. However, this comparison abstracts from a more complicated interaction of various determinants of solvency regimes. The actual degree of “stringency” of capital requirements for insurance activities within a given solvency regime depends on the confluence of valuation standards, the definition of capital, the level of solvency thresholds (that is, prescribed capital requirement and the minimum capital requirement), and the sensitivity of capital requirements to the changing nature of risk drivers and interaction (“diversification effects”) as well as the scope and implementation of supervisory practices.1Internal models and additional reporting by legal entities can provide the basis for additional insight on group-wide activities, especially in cases when the legal entity accounts for the majority of group-wide activities.

Figure 17.5b Overview of Solvency Regimes: Valuation Standards



wide vulnerabilities gained from routine supervisory report-ing.40,41 Stress test scenarios may also introduce specific risks (which are not addressed in the relevant solvency regime) and/or augment the severity of risk factors that already exist in the relevant solvency regime.42

The scenario design is driven by the relevant macro- financial vulnerabilities to identified risk factors and their implications for individual and system- wide solvency. This requires modeling different types of shocks to risk factors and their transmission channels through the linkages between insurance activities and the rest of the financial system as well as the real economy (Figure 17.6). Scenarios should be plausible and meaningful in relation to the capacity of firms to control and mitigate vulnerabilities to chosen risk factors. Considering the mitigating impact of capital planning and expected strategic changes allow for demonstrable anticipa-tion and integration of findings in current processes (“use test”) (Jobst 2013a).

Risk scenarios ideally combine historical and simulated (or hypothetical) outcomes subject to supervisory judgment and/or expert opinion. This approach helps avoid optimiz-ing the calibration of adverse scenarios based on historical experience without recognizing the aberration of risk fac-tors that might result in different scenarios in the future.43 Even the worst case scenarios would need to satisfy a gen-eral concept of plausibility. Such plausibility would ideally be defined as a probabilistic concept, which deems a sce-nario more plausible the higher the probability of realiza-tion consistent with the prevailing correlation of risk factors (Breuer and others 2012).44

It is important to cover all relevant risk categories. Sce-narios commonly include a combination of market and credit risks (from both corporate and sovereign exposures), interest rate risk from asset- liability mismatches, foreign currency risk, liquidity risks, underwriting risks, and concentration risks,

valuation standard (Figure 17.5b).38 Inflating discount rates (by a certain credit spread and/or liquidity risk premium in addition to the risk- free rate) for the estimation of technical provisions under a less market- sensitive approach could over-state solvency ratios (and increase liquidity risks ex ante). In-surers would have the incentive to invest in as illiquid and risky assets as possible to maximize returns while at the same time minimizing technical provisions under less risk- sensitive valuation. Thus, adjustments to statutory data might be re-quired in stress tests if lower (or no) market consistency within a given solvency regime were to dominate its sensitiv-ity to risk factors under different stress scenarios. Removing mitigating factors and alleviations tends to generate lower capital resources (through recognizing economic loss) and in-crease capital requirements more than under a cost- based valuation standard.

Given their considerable reliance on prudential data, the valuation standards in FSAP insurance stress tests are heav-ily influenced by the existing (national) solvency regimes and may vary significantly across countries. Most supervi-sory stress tests tend to be based on statutory accounting (us-ing historical cost), which limits a wider application of a more risk- sensitive valuation in the context of FSAPs. A fully market- consistent valuation has been applied only to a few countries (Belgium [2013], Canada [2014], Portugal [2007], Spain [2006], Singapore [2013], and Switzerland), with some adjustments for unrealized gains and losses in other cases (Japan [2012]).39 The most comprehensive set of valua-tion approaches in FSAPs to date was used for Belgium (2013), which included both statutory (cost) accounting, a “near market- consistent” valuation according to EIOPA’s Quantitative Impact Study 5 on Solvency II, and a fully market- consistent valuation of both assets and liabilities. Generally, the implementation of Solvency II implied a sig-nificantly more market- consistent valuation basis in recent FSAPs in EU and the European Economic Area member countries (compared to FSAPs in other countries).

Scenario Design and Other Assumptions

The assessment of capital adequacy under stress requires the definition of severe but plausible macro- financial scenarios. A stress test should be designed to quantify the implications of a rapid deterioration in earnings and/or capital and re-serves due to adverse changes in one or more risk factors af-fecting both investment and underwriting performance. Findings from a wide range of loss scenarios within a stress test can complement the insights on firm- specific and system-

38 For instance, the Solvency II regime contains some dampeners and countercyclical elements— the volatility adjustment, matching adjust-ment, convergence period for extrapolating the basic risk- free curve, which are aimed at increasing the robustness of the solvency regime but also imply strong assumptions.

39 The reference to certain years only applies to countries that have com-pleted two insurance stress tests in the context of an FSAP exercise.

40 Note that an adverse scenario does not imply an adverse development of all risk factors. Some asset classes (such as sovereign bonds in major juris-dictions) might appreciate during times of stress due to negative selection in asset markets (“safe havens”).

41 It would, however, be impossible to map all vulnerabilities for all busi-ness models in all jurisdictions without running the risk of obscuring the most relevant macro- financial risk transmission channels affecting financial stability.

42 For example, if the solvency framework does not consider sovereign risk, the scenario can be adjusted. In addition, if the current valuation does not require insurers to recognize unrealized losses from sovereign bond investments, the valuation needs to be adjusted accordingly.

43 In addition, past occurrences themselves might have limited explana-tory power as early warning signals (“legacy/hindsight bias”).

44 More specifically, this would amount to the first- order stochastic domi-nance of the selected stress scenario over the outcomes of alternative combinations of risk factors (Abdymomunov, Blei, and Ergashev 2011). All risk measures within the plausible domain of outcomes would need to satisfy the axioms of coherence (Artzner and others 2001). A risk measure is deemed “incoherent” if it violates the axioms of subadditiv-ity, monotonicity, positive homogeneity, and translation invariance. For example, subadditivity, which is a mathematical way to say that diversi-fication leads to less risk, is not satisfied by value at risk.



While macro- financial linkages of underwriting activity are generally limited, there can be tail dependence between market and insurance risks during times of stress. For in-stance, large catastrophes could have negative effects on as-set prices and/or funding conditions affecting insurance companies. Rising risk aversion and a higher cost of capital during an economic downturn could accentuate these ad-verse conditions, raising the cost of replenishing capital after large insurance losses.45 Also, longer term uncertainty about the ultimate consequences of a catastrophe (or a pandemic like the Severe Acute Respiratory Syndrome outbreak in 2002/03) tends to be negative for the stock markets. Thus, amending a catastrophe scenario by some subsequent market stress (especially a stock price decline and potentially some currency fluctuations) could be considered. Moreover, higher rates of inflation during periods of economic recovery can adversely affect provisioning and reserve adequacy in non- life underwriting, especially if changes in claims activ-ity negatively impact investment performance in real terms. Also some NTNI activities, such as funding and hedging arrangements via capital markets (for instance, securities fi-nancing transactions and over- the- counter derivatives), dif-fer markedly from long- term cash- flow projections associated with insurance liabilities and are inherently more susceptible to cyclical effects than is the mainstream insurance business (IAIS 2013c).

including the interconnectedness with other financial institu-tions. When selecting relevant risk categories, it is important to consider the main features of insurance companies as un-derwriters and investors, such as the share of traditional busi-ness versus unit- linked business, average guaranteed rates, and modified duration of assets and liabilities as well as the degree of interconnectedness within the industry and other financial sectors.

The scenario design will also need to reflect the character-istics of the insurance market, which comprises a wide range of business models and supervisory frameworks that differ significantly across jurisdictions. As a result, the relevance of macro- financial shocks and their balance sheet impact are bound to differ from those that apply to other financial insti-tutions and insurance companies in other countries. The ap-plication of several scenarios helps cover all relevant risk categories and generate stress test results with greater vari-ability regarding future outcomes. Typical scenarios include one or more of the following risk factors and macro- financial transmission channels (Box 17.3):

• Recession: Decline in equity and property prices, in-crease in credit spreads, higher lapse rates, and higher default risk of mortgage borrowers, offset by lower interest rates

• Financial sector and/or sovereign debt crisis: Higher credit spreads for financials, default of one or more large bank counterparties, and depreciation of all as-set classes, including sovereign bonds

• Inflation scenario: Claims inflation and lower real in-terest rates, offset by rising equity and property prices

• Non- life underwriting shock: Large catastrophe claim(s) due to natural hazards or man- made disaster, which could be combined with a default of a reinsurer or a decline in equity prices (in the country affected by the catastrophe)

• Life underwriting shock: High lapse rates and/or a pandemic

45 However, recent studies show that even highly disastrous events like Hurricanes Katrina and Sandy in the United States or the 2011 Tōhoku earthquake and tsunami in Japan had only very short- lived effects on the respective domestic stock markets (Wang and Kutan 2013), and broad market indices did not decline significantly. This could be explained by the fact that the economic loss in terms of GDP was rather limited. In emerging and developing countries, a natural catastrophe might have a much more severe impact on the real econ-omy and the domestic financial market, especially if large parts of a country or a large metropolitan area with critical infrastructure are affected.

Sources: PRA 2013; IMF staff; and authors.Note: FSAP = Financial Sector Assessment Program.1Standards assessment of insurance core principles, which considers the prudential oversight of risk management and governance aspects.

Figure 17.6 Elements of Risk Assessment and Scope of FSAP Stress Testing

Risk Drivers(investment

andunderwriting)

Scope(general orspecific to

business lines)

BusinessRisk

ExternalFactors

ManagementBehavior andGovernance

RiskManagementand Controls

Capital andSurplus Liquidity Resolvability Substitutability

Yes Yes – – Yes Yes – –Indirect via standardsassessments1

Coverage in macroprudential stress testing of insurance sector in IMF FSAPs

Gross Risk Risk Mitigation

Potential Impact Risk Context Operational Financial Structural



46 Single- risk factor shocks can be calibrated based on time series analysis (for instance, by means of bootstrapping procedures).

47 Extending the time horizon would then require the inclusion of man-agement actions and changes in policyholder behavior; also the pricing cycle in the (re)insurance market would need to be taken into account.

zon constitutes a fundamental change in the FSAP stress testing methodology and calls for making several additional assumptions when designing the scenarios.

Despite the prevalence of single- period stress testing ap-proaches for insurance, there are clear advantages associated with multiperiod scenarios. Extending the stress test horizon to multiple periods helps identify medium- and long- term vulnerabilities from a gradual erosion of the solvency position, which would inform remedial actions and recovery plans (but also support capital planning decisions).47 It also allows for a more comprehensive coverage of intertemporal effects of shocks (such as the impact of lower solvency/rating down-grades on the scenario- based cost of funding/underwriting capacity) and mitigating factors (such as the impact of de-ferred tax assets, dividend policy, and managerial actions).

While the application of a longer time horizon better reflects the long- term nature of most underwriting activi-ties (except for “ short- tailed” non- life insurance), it also di-minishes the accuracy of any forecast of solvency conditions under stress. Moreover, the effectiveness of management ac-tions, like changes in hedging activities, product design, and dividend payouts, are difficult to model and compare across firms, which risks undermining the consistent implementa-tion of a stress testing methodology. FSAP exercises tend to abstract from a quantitative assessment of these mitigating factors but recognize the scope available to insurance man-agers for allocating losses among current and future benefits and equity (as in the case of France [2005]) (Figure 17.6). Given their deeper understanding of local markets and firm characteristics, supervisors are better placed to assess the credibility of management actions and their mitigating ef-fects under stressed market conditions in multi period stress tests.

The scenario specification starts from a baseline that re-flects expected changes in risk factors without shocks to macro- financial conditions. Forecasts for some risk factors (such as interest rates, credit spreads, and real estate prices) could be directly derived from macroeconomic models, while other factors would likely be estimated exogenously (or be based on expert judgment). The adverse scenarios should be defined as deviations from the baseline scenario at suffi-ciently high but realistic statistical confidence for the cali-bration of the magnitude of shocks. Traditionally, most FSAP stress test exercises (much like similar tests completed by country authorities) apply single- factor shocks. However, these shocks are individually determined rather than cali-brated jointly using the historical sensitivity of investment and underwriting performances to changes in macroeco-nomic conditions.46

Single- period stresses are the predominant modeling framework used by national supervisory authorities. In most cases, risk factors are applied as one- off (instantaneous) shocks that are exogenously determined or calibrated to a specified statistical confidence level over a one- year risk horizon. Au-thorities in Canada, Singapore, and the United States apply multiyear period scenarios with projection horizons of up to five years, while others are using single- period or instanta-neous shocks. FSAP stress tests were also focused on a single- period stress until very recently. However, the actual concept differed slightly in various countries. In some cases, a one- year horizon was used with the shock occurring at the end of this year, while in other exercises an instantaneous shock was applied. The only exceptions are three stress test exer-cises in Japan (2012), Singapore (2013), Canada, Denmark (2014), and South Africa (2014), with two-, three-, or five- year projection horizons. This extension to a multiyear hori-

Box 17.3. Recessionary Scenarios in Insurance Sector Stress Testing

A comprehensive stress testing approach would need to consider the economic impact of a recession not only on the asset side of an in-surer’s balance sheet but also to insurance liabilities. There are several transmission channels between adverse changes in economic condi-tions and the performance of insurance companies:

• Lower household wealth and disposable income could lower the premium income of insurers (and increase the volume of claims). Life insur-ance, but also motor insurance, tends to be more sensitive to a downturn of the economy (EIOPA 2013b). Dionne and Wang 2013 identify a business cycle pattern in fraudulent insurance claims for automobile theft in Taiwan POC. Higher claims in non- life insur-ance could also be attributable to higher operational risk.

• Higher household wealth and disposable income could increase lapse rates in life insurance. The lapse risk would depend on the incentives of policyholders to surrender their policies; if the surrender value is high (and/or the interest rate levels rise above the implied return from guaranteed term insurance), the lapse rate tends to increase.

• Higher corporate default rates could increase claims from financial guarantees and credit insurance. However, these lines of business are quite heterogeneous, ranging from trade and export financing to mortgage insurance and credit enhancements of structured fi-nance transactions. All these subcategories would require a specific set of assumptions regarding probability of default and loss given default.



property, forestry, and project development). The shocks could be applied to both investment assets and self- used property.

• Foreign exchange: Risks inherent to foreign exchange rate fluctuations are usually seen as less relevant since many solvency regimes include strict matching rules for underwriting and investment activities denomi-nated in foreign currencies. Instead of providing sev-eral shocks for bilateral exchange rates, sometimes a general appreciation or depreciation of the local cur-rency could be adequate.49

• Credit risk: Credit risk can be highly relevant for in-surers due not only to investment holdings of fixed income instruments. Credit risk is also inherent in derivatives transactions, contractual relations with reinsurers, and direct lending where allowed in the relevant legal framework.50

• Concentration risk: Concentration risks can be preva-lent on both the asset and the liability sides of insur-ers’ balance sheets. For a consistent measurement of this risk, it is useful to apply the stress to the com-bined exposure to a single counterparty (for exam-ple, a reinsurer to whom business is ceded while an undertaking also holds asset exposures to the same entity). Also, concentrated banking exposures could be stressed by assuming the default of the largest bank counterparty and modeling the effects by tak-ing into account the different levels of seniority of deposits, secured or unsecured bonds, equity, repo lending, or over- the- counter derivatives transactions; if the bank were to act as a distribution channel for insurance products, lower premiums could also be included in the stress scenario.

• Liquidity/funding risk: Insurers invest premium in-come from long- dated gross claims and life insurance provisions in high- quality, liquid assets to support (mostly predictable) short- term payment obligations from insurance policies. However, an abrupt rise in the frequency and severity of claims (due to an excep-tional string of large natural catastrophes) could

Risk Factors and Aggregation Approaches

The relevance of risk factors can vary depending on the prev-alent business models and products as well as the common investments of insurance companies. In general, the most significant macro- financial risk transmission can be found in forward- looking indicators of monetary conditions (inter-est rates and inflation) and asset valuations in capital mar-kets (equity and debt prices) that affect the performance of insurance companies. Although income from underwriting activities is the dominant driver of earnings, life insurers also depend on their investment performance, which can be im-pacted adversely by interest rate changes and asset price vola-tility, especially if some investments are very long term and/or are highly concentrated in certain asset classes. In addi-tion, non- life insurers with long- term claims are sensitive to significant changes in inflation, which affects their loss provisioning.

• Interest rates: Interest rate risk is one of the most im-portant risk factors, especially for life insurers offer-ing long- term annuities with guarantees, since the duration of their assets is usually shorter than the duration of liabilities. The methodological approaches for generating interest rate shocks vary widely and include simple parallel shifts of the interest rate term structure as well as more advanced modeling in line with macroeconomic projections. In the case of a re-cessionary scenario, interest rates would likely de-cline or remain at a low level given the (expected) accommodative monetary policy. Inflationary pres-sure, however, would likely result in a scenario with upward- moving interest rates. Generally, short- term interest rates tend to be more volatile than long- term rates (Box 17.4).48

• Equity: Equity risk is a typical component of insur-ance stress tests although the relevance of equity ex-posures has decreased in many countries over the last decade. The main challenge is to determine shocks for very different categories of equity exposures, ranging from listed stocks to private equity, hedge funds, and various other alternative asset classes. Similarly, for strategic participations, shocks can be designed to reflect the persistence of these invest-ments (and its impact on their adequate valuation).

• Real estate: Property price stresses can be designed in the same manner as equity shocks; this asset class is characterized by a high degree of heterogeneity across types of real estate (residential/commercial

48 Interest rate risk can be decomposed into changes in the short-term risk- free rate, term premium, and counterparty risk (which also includes changes in sovereign risk). Thus, the impact of interest rate shocks is commonly modeled as the combined result of valuation changes in in-terest rate- sensitive assets and the losses associated with valuation hair-cuts due to increases in credit spreads.

49 Nevertheless, in a realistic adverse scenario, some currencies are likely to appreciate due to a “flight to safety.”

50 From a modeling perspective, there are two interrelated ways of stress-ing credit risk exposures of insurance undertakings: First, credit risk in a narrow sense, or counterparty default risk, can be modeled similarly to common practices in banking stress tests by estimating stressed prob-abilities of default and losses given default for different types of claims; however, the historic evidence of defaults for relevant types of claims (especially for reinsurance defaults) is scarce, so some approximations are needed. Second, market prices of bonds or other fixed income in-struments could be stressed by assuming higher credit spreads. Depend-ing on the exact scenario, different shocks might be assumed for corporate bonds (even distinguishing between financials and nonfinancials, or tak-ing different seniority levels into account) and sovereign bonds. Deriving a consistent macroscenario for sovereign spread should also include a po-tential “flight to safety” effect, similar to foreign exchange shocks.



• Contagion risk: Some arrangements between bank-ing and insurance activities are prone to contagion effects during times of severe stress but are difficult to model and/measure.52 Example include committed funding arrangements and contingent intragroup transactions. Against a backdrop of an overall loss in confidence in capital markets, the banking side of conglomerates (or bank counterparties to liquidity swaps) could become vulnerable to the risk of large withdrawals of deposits and/or the runoff of liabilities. As both banks and insurers would sustain

drain the existing liquidity position and overwhelm the liquidity management capacity of non- life insur-ers.51 Liquidity risks could also materialize for life in-surers, though usually to a lesser extent, if payment obligations rise above actuarial expectations due to structural changes in claims activity and/or negative cash flows from exceptional surrender behavior by policyholders (“lapse risk”) (Box 17.5). For instance, unexpected surrender payments due to higher lapse rates would require insurers to use cash reserves or sell assets to meet these obligations.

51 Note, however, that the immediacy of such liquidity pressures is quite distinct from the demands placed on the treasury function of banks in wholesale funding markets (Box 17.4), where margin calls have to be satisfied on an intraday basis.

52 Conglomerates could also engage in liquidity transformation between the insurance and banking entities if liquid assets were transferred to the banking entities in exchange for less liquid assets. This would allow the banking part of the conglomerate to satisfy liquidity requirements, while the insurer would benefit from higher asset returns.

Box 17.4. Assessing the Impact of Low Interest Rates on Insurance Activities

Since insurers are large investors in fixed income instruments, equity, and real estate, they are particularly vulnerable to interest rates (and the associated reassessment of term and inflation risk premia). Unlike banks, which benefit from lower short- term interest rates (which lower borrowing costs) and the likely widening of term spreads, the opposite is true for insurance companies. Low rates increase the insur-ers’ long- term liabilities in today’s terms. In most cases (except for most non- life insurance business lines), the duration of these liabilities exceeds that of available investment assets. On the asset side, low interest rates reduce investment returns and increase the reinvestment risk of assets. This challenge of a “ short- long duration mismatch” in a low interest rate environment is even more pronounced for firms that need to match long- term, low- risk investments to guaranteed rates of returns to policyholders. If the duration of liabilities exceeds that of assets, and interest rates decline, lower investment income increases the insurer’s dependence on underwriting performance and/or could encourage greater risk taking (once gains from higher yielding assets have been realized).1

Adverse effects from low interest rates vary by the balance sheet structure and the type of business. The interest rate risk of existing poli-cies (that is, the legacy book) in life insurance can be significant, as future premiums cannot be adjusted to reflect lower investment returns, and the higher value of interest- dependent assets usually can not compensate for the higher present value of liabilities due to the negative duration gap. Low interest rates would require insurers to either increase premiums for the same expected future claims payments or lower guarantees to policyholders lest they risk reducing future earnings. While there are usually no tight substitutes for insurance, and setting higher premium rates should be theoretically possible, in practice, insurers would be reluctant to change their pricing conditional on investment returns.

Both life and non- life insurers would need to take lower investment returns into account in the pricing of new underwriting. However, low interest rates are unlikely to cause a solvency impact on non- life business in the absence of negative demand effects and lower ex-penses due to low inflation expectations. Similarly, some life insurance products (mortality, disability, and long- term care) have more pro-tection features than saving features. These protection- oriented features would allow insurers to compensate lower investment returns with higher risk charges. However, demand for those less vulnerable businesses (that is, non-life and protection-oriented life products) is still inherently susceptible to economic conditions and is likely to decline during recessions (and lower interest rates).

Insurance supervisors have identified the low- rate environment as a major risk for the life insurance industry (EIOPA 2013a; Antolin, Schich, and Yermo 2011; Swiss Re 2012). However, the quantification of the capital impact from low interest rates is not straightforward. Stress tests that use an instantaneous interest rate shock without market- consistent valuation cannot capture the long- term effects on solvency. Nevertheless, some methods that can provide rough estimates have been presented in recent years. French and others 2011 project cash flows based on the existing investment portfolio and the duration of insurance liabilities to determine the sensitivity of own funds to changes in markets rates. They assume that maturing bonds would be reinvested at a lower market rate and that asset allocation remains unchanged. In its 2013 Financial Stability Review, the Deutsche Bundesbank 2013 used scenario analysis to examine the effect of lower future investment income on bonus and rebate provisions over a 10-year horizon with a view to drawing conclusions about the sol-vency ratios of 85 German life insurers. The analysis was based on a refinement of the model developed by Kablau and Wedow 2012, who found that in “a stress scenario with a prolonged period of low interest rates, more than one- third of German life insurers would no longer be able to fulfill the regulatory own funds requirements under the current solvency regime (Solvency I) by 2023. […] This result is attribut-able primarily to high guaranteed interest rates” (Deutsche Bundesbank 2013, 69).

1 This situation is potentially aggravated by a higher substitutability of some life insurance products and negative demand effects impacting pre-mium income from life insurance due to lapse risk.



to any disputed claims, particularly where the out-come is subject to legal proceedings); the impact of inflation; the effects of increasing longevity on pen-sion products; the guarantees and options in policy terms; the risks of early policy surrenders (“lapse risk”), which can be linked to interest rate changes as well as other socioeconomic, legislative, and techno-logical changes.

• Demographic risks: Changes in long- term trends of mortality can have a significant permanent impact on the life insurance industry. While shocks do not change the underlying trend, they heavily affect both the level and volatility of mortality rates and long- term payouts.

• Catastrophe risks: This risk reflects the capacity of in-surers to absorb higher actual claims and/or unexpected exposures or the aggregation of claims associated with catastrophic events, or the possible exhaustion of reinsurance (or alternative risk- transfer) arrange-ments, and the appropriateness of the underlying as-sumptions and calibrations underpinning catastrophe models. Insurance companies use commercially available models to estimate the possible cost of claims arising from natural catastrophes and man- made disasters. The models are based on historic claims and are constantly updated. Nevertheless, there is a fundamental model risk, especially for low- probability events.

The definition of risk categories should provide additional insights into the stability of the insurance sector under stress outside the existing solvency regime (Box 17.6). It is not suf-ficient to design a stress test exclusively along the lines of prudential measures, which might be limited to general vul-

a sharp decrease in the value of their investment portfolios, funding needs could lead to greater reli-ance on intragroup transactions (or the use of con-tingent funding arrangements).

Moreover, there are insurance- specific risks (IAIS 2003):53

• Underwriting (or premium) risk: Commercial consid-erations regarding the pricing and coverage of insur-able interest are influenced by the rapid changes in the volume of the underwriting portfolio, uncer-tainty of the claims experience (for example, the vol-ume and timing of claims), and tolerance for variations in expenses. Moreover, the dependence on intermediaries (such as brokers and securities under-writers), the possibility of higher reinsurance rates, and the effects of high pricing uncertainty in new business lines or underwriting activity in emerging market and developing economies (possibly compli-cated by insufficiently understood insurance risk and reserving requirements) represent considerable chal-lenges to the risk management of insurers.

• Reserve adequacy: This includes the declining reserve coverage through technical provisions; the uncer-tainty of the claims experience (in terms of the fre-quency and size of claims); the length of the claims development (including possible outcomes relating

53 The inclusion of underwriting risks is essential; however, not all underwriting risks are equally relevant. Fewer risk types are likely to be included in stress tests covering shorter time horizons. In this case, a stress test could include an instantaneous shock due to a large catastrophic event (natural or a man- made) and/or a mass lapse event, but probably no improvement in longevity, which is a more gradual development.

Box 17.5. Liquidity Risk in Insurance

Rising liquidity risk tends to amplify the deterioration of a firm’s capital position under adverse scenarios and should be considered an es-sential element of an insurance stress test.

In general, the long- term funding profile of insurers is less susceptible to funding shocks than banks (although such risks cannot be ex-cluded). However, insurance companies may still have liquidity and maturity mismatches, and the duration gap tends to be negative (espe-cially for life insurers). Moreover, some financial transactions, such as over- the- counter derivatives for hedging and securities financing transactions, are subject to capital market conditions and could create short- term cash- flow needs (requiring the availability of high- quality, liquid collateral) that are markedly different from the long- term cash- flow projections associated with insurance liabilities. In many countries, insurance regulations limit liquidity risks through restrictions on illiquid investment (such as loans or real estate) as well as the prohibition of certain derivatives and securities financing transactions.

Stress testing liquidity risk is most relevant for non- life insurance and reinsurance. In many countries, insurance supervisors monitor li-quidity positions of reinsurers and non- life insurers by comparing their liquid assets with potential payment amounts associated with large claims. Liquidity stress tests can flag specific vulnerabilities faced by reinsurers that would have to settle large claims after a major natural catastrophe.1 However, there is no well- established market practice of liquidity stress within the industry yet. One possible approach is to make use of cash- flow projections with certain stress scenarios (such as large claims from catastrophe events, lower future premiums from commercial lines in response to greater competition, and collateral needs from over- the- counter derivatives transactions).

1 This is also relevant for life insurers experiencing a significant increase in surrender rates.



cover sovereign risks, but the actual modeling differs sig-nificantly.54 Risk factors on the liability side vary among jurisdictions and are less comprehensive than those on the asset side. More than half of the stress tests consider some form of life underwriting risk (mortality, morbidity, and longevity); however, other material risks affecting life insur-ance companies, such as pandemics and lapse/surrender risks, are rarely included. For non- life underwriters, super-visors acknowledge economic losses from man- made and natural disasters as peak risks based on the maximum his-torical claims experience and/or the aggregate policy limit. Other non- life risks tend to be focused on the relative

nerabilities for less extreme changes in risk factors. In addi-tion, the ability to vary the severity of shocks at different degrees of statistical confidence highlights the sensitivities of the capital contingent on the dynamics.

While supervisory stress testing approaches often in-clude a comprehensive set of asset price shocks, the coverage of underwriting risks varies significantly across countries (Table 17.1 and Appendix Table 17.1.4). Most of the exer-cises include all material asset risks (credit, market, and in-terest rate risks), which sometimes exceed the scope of the existing solvency regime. However, foreign exchange and sovereign risks as well as other risks from the deterioration of the insurer’s risk profile and second- order effects of shocks affecting future underwriting capacity and diversifi-cation benefits are rarely found. Foreign exchange risks are explicitly included only in stress tests completed by national authorities in Bermuda, Czech Republic, Guernsey, Swit-zerland, and the United States. Some jurisdictions also

Box 17.6. Examples of Supervisory Approaches of Insurance Stress Testing

Several jurisdictions have made sustained efforts in developing comprehensive stress testing frameworks, which are largely based on bottom- up approaches that involve considerable involvement by insurers (and their own risk models). These jurisdictions include (Appen-dix Table 17.1.2):

European Union: The Committee of European Insurance and Occupational Pension Supervisors and its successor, the European Insur-ance and Occupational Pension Authority (EIOPA), completed their first EU- wide stress tests in 2009 and 2011, respectively.1 Both stress test exercises used a (near) market- consistent valuation of assets and liabilities under the Solvency II framework (Figures 17.5a and 17.5b). While the scope of the first stress test included only 30 large insurance groups, the scope was expanded in the 2011 exercise to include additional insurance groups and solo entities. Overall, the stress test in 2011 covered more than 50 percent of the European insurance sector in terms of balance sheet assets. Three scenarios were designed for the 2011 stress test: (1) a baseline scenario with slightly negative capital market developments, (2) an adverse scenario with more pronounced equity, property, and credit- spread shocks, and (3) an inflationary scenario with sharply rising interest rates. Some underwriting risks were included, for example, an increase in longevity, a natural catastrophe (with companies providing their individual largest maximum probable loss for a 1- in- 200-years event), a claims deficiency shock, and an increase in lapses. Besides these three scenarios, two satellite exercises were included to cover sovereign stresses (modeled via an increase of sover-eign bond spreads) and a prolonged low- yield environment. No specific confidence level was provided for the shocks, which were sup-posed to happen instantaneously, thereby ruling out any discretionary ex post management actions. As an output measure the minimum capital requirement as well as the available capital had to be submitted by participating companies. EIOPA published aggregated results, but no individual company data (EIOPA 2011a).

Singapore: The Monetary Authority of Singapore (MAS) conducts comprehensive stress testing exercises covering all direct insurers. Several exercises have been conducted with different time horizons and time dynamics ( short- term [1 year], medium- term [3 years], and stress- to- failure scenarios). Short- term and medium- term scenarios are specified by the MAS, while stress- to- failure scenarios are devel-oped by the appointed actuary of each participating insurance company. The shock scenarios comprise a rise in mortality/morbidity, changes in yield curves, an equity market crash, higher operating expenses, a decline in new underwriting, higher lapse rates, and other risk factors that the appointed actuary considers as relevant. The stress test report submitted to the MAS breaks down the contribution of each risk driver to the overall economic impact of all shocks and the mitigating effect of potential management actions. The report also includes recommendations on risk- mitigating actions by the appointed actuary. In addition, the MAS requires the Board of Directors of each insurer to discuss the results and recommendations by the appointed actuary, comment on the feasibility of the management ac-tions, and conclude whether any measures need to be taken based on the findings from the stress test.

Canada: The Office of the Superintendent of Financial Institutions (OSFI) introduced a guideline on stress testing in 2009, which covers the purpose, role of the Board of Directors and senior management, methodology, and scenario selection, and more specific guidance on risk mitigation, securitization, reputation, counterparty, and concentration. The guideline covers not only insurers but also their holding companies, banks, and bank holding companies, and, thus, provides an integrated framework for the entire financial sector. In 2012, OSFI requested several life (re)insurers to complete a macroeconomic stress test based on a common adverse scenario. OSFI shared the indi-vidual results with the participating companies.

1 By EU Regulation, EIOPA is obliged to perform stress tests on a regular basis.

54 For example, the joint exercise by EIOPA and Switzerland includes sov-ereign risks mainly in the form of higher spreads (without considering explicit default scenarios). However, the actual severity of shocks varies considerably among jurisdictions.



and downward shocks were tested; however, the low interest rate environment motivated greater inci-dence of upward shocks since 2011 (Japan [2012], Luxembourg, Norway, Singapore, and South Africa [2014]), at least in the form of separate sensitivity analyses (Denmark [2014]). Some stress tests applied more sophisticated variations of the interest rate term structure, like a steepening or a flattening of the yield curve (Belgium [2013], France [2005], Sin-gapore [2013], Spain, and Switzerland). Very few stress tests applied different shocks for domestic and foreign interest rates, such as in the cases of Mexico and Denmark (2014).

• Equity shocks were rather homogeneously modeled. For most FSAP stress tests, a uniform shock between 25 and 35 percent was assumed without considering differences in industry sectors or specifying diver-gent shocks for equity- like assets other than shares like hedge funds or private equity.57 Smaller shocks were applied if equity markets were already de-pressed (Japan [2012]) or after significant financial sector transformation (Belgium [2013]).

• Property price shocks were applied uniformly (similar to equity shocks). In most FSAP stress tests, real estate prices declined by between 15 and 30 percent during times of stress. However, most exercises did not (1) differentiate between commercial and residential real estate prices or (2) involve different shocks to real estate exposures in other countries. Also, changes in the collateral value of mortgage loans were not mod-eled explicitly.

• The specification of credit risk varied significantly. Un-til 2007, one approach to representing a credit risk shock was to use the default rates implied by rating categories under the standardized approach for credit risk of the Basel II framework for banks, where unexpected losses are defined as the product of the credit risk weight and the minimum capital require-ment. This approach was used in FSAPs for Spain and Portugal. Simple valuation haircuts were an-other rather straightforward way, like a 1.5 percent loan loss in Japan (2003) or a 4.4 (9.4) percent loss on loans (corporate bonds) in the case of Israel. Downgrade scenarios for bond holdings (for exam-ple, two to four notches) were frequent during the early phase of the financial crisis (Guernsey and the Isle of Man), whereas absolute (for example, an in-crease by 50 basis points) or relative (for example, multiplying current spreads with a factor of 1.5) shocks to credit spreads have now become the norm in stress testing credit risk from traded securities. A differentiation of spread increases by rating class was

impact of premium risk and the rising cost of insurance claims; however, the counterparty risk of a defaulting rein-surer and the basis risk in reinsurance programs and alter-native risk transfer are frequently not covered.55 Most exercises do not explicitly consider shocks of off- balance sheet exposures, with the exception of Bermuda and the Czech Republic.56 However, supervisors who approve inter-nal models have tended to extend their methodologies to incorporate firm- specific worst- case scenarios based on in-ternal models and/or scenarios that combine the impacts of both adverse macro- financial conditions and maximum ag-gregate underwriting losses.

The scope of risk factors in FSAP stress tests is generally more limited compared to that of supervisory exercises (Ta-ble 17.1 and Appendix Table 17.1.3). As much as FSAP in-surance stress tests benefit from a close cooperation with national supervisors in collecting essential data on insur-ance risks (which are mostly assessed via BU approaches), the reliance on existing stress testing frameworks also limits the extent to which other risks can be analyzed (and com-pared across FSAPs for different countries). In most cases, alternative specifications of economic shocks are often con-fined to sensitivity analyses, which are combined with an overall macroeconomic scenario that primarily impacts the investment performance of insurers. Thus, FSAP stress tests tend to be biased toward risk factors affecting investment performance, such as equity and interest rate shocks (and, to a lesser extent, real estate and credit- spread shocks), which are common to most supervisory approaches. How-ever, the combined effect of economic and underwriting shocks, feedback effects, and the sensitivity of stress test re-sults to changes in the aggregation of risk factors are fre-quently outside the scope of FSAP insurance stress tests, which also tend not to incorporate management actions in a more dynamic capital assessment under stress (Figure 17.6). More specifically, risk factors in FSAPs were specified as follows:

• Interest rate shocks varied substantially in terms of se-verity and implementation. The standard approach of a parallel shift of the yield curve was used in most countries with shocks usually between 100 and 250 basis points. In most of the exercises, both upward

55 The market for alternative risk transfer instruments has grown consider-ably as insurers are expanding their business activities to capitalize on fee income and satisfy demand for cost- efficient (re)insurance capacity. Some of the approaches adopted by firms include expanding their asset management services for sophisticated investors, adopting alternative collateral management solutions, and/or establishing so- called “side-cars” as well as creating special- purpose insurers or segregated accounts companies. Most specialized insurers issue insurance- linked securities, which have become the hallmark of an evolving alternative risk- transfer market (BMA 2013c).

56 For example, the BMA requires an estimation of the impact of a two- notch downgrade of the counterparties. Exercises of other jurisdictions cover at least partially some off- balance sheet risks (such as counter-party credit risks of over- the- counter derivatives) within their overall credit risk module.

57 Different shocks were applied for advanced economies and emerging economies only in the FSAPs for Belgium (2013), the Netherlands, and Singapore.



ing technical result and higher operating costs (Bel-gium [2006] and the Netherlands).

• Other risk factors. In the FSAP for the Netherlands, a commodity shock was added as well as an increase in implied volatilities (which was also used in the South African FSAP, 2008).

The overall impact of these single- factor shocks on the balance sheet of insurers is determined by some form of ag-gregation to account for their dependence. Many FSAP in-surance stress tests used a simple summation (Denmark [2014], Guernsey, Isle of Man, Japan [2012], Singapore [2013], South Africa, and the United States) while others added up certain combinations of single- factor shocks (Ber-muda, France [2012], and Mexico). In some cases, the total impact was derived from aggregating the individual impacts of various risk factors using one or more correlation matrices (which is also embedded in several solvency regimes, most notably Solvency II). Other countries used aggregation ap-proaches via correlation matrices among risk factors (Bel-gium [2013], Germany [2016], Luxembourg, Netherlands, Norway, Portugal, and Switzerland). However, this approach implies that the changes of the risk factors are random with a given correlation, which is inconsistent with the notion that shocks represent a significant deviation from expectations around a random process. So the correlation- based approach was sometimes complemented by simple summation, that is, setting correlation coefficients to one. Thus, a more intuitive approach would be to consider the total impact based on a linear combination of the separate risk- factor impacts, which preserves individual risk- factor impacts at high levels of sta-tistical confidence on an aggregate basis. Some supervisory stress tests (as well as FSAP exercises) have adopted a dual approach of assessing capital adequacy under stress with and without the aggregation of risk factors with diversification ef-fects (Box 17.7). In some FSAPs, especially those prior to the global financial crisis, the impact of risk factors was not ag-gregated (for example, Belgium [2006], Spain, and Denmark [2007]).

Output Measures

The main output variable in all solvency stress tests is the change in capital adequacy due to the individual or joint im-pact of one or more single- factor shocks and/or scenarios over a given forecast horizon. Under the total balance sheet approach, which underpins insurance solvency regimes in many countries, insurers would need to maintain a positive net asset value with a high level of statistical confidence, usually over a one- year risk horizon, subject to risk factors impacting the value of assets, the sources of funding, and the payout of insurance claims. The solvency ratio of an in-surer is calculated as the excess of capital (assets minus liabil-ities, usually with some restrictions to account for the quality of different types of equity capital, such as preferred stocks) over the prescribed capital requirement or some other na-tional capital standard. Besides the prescribed capital

used in the Netherlands as well as in many exercises like Belgium (2013), Denmark (2014), Germany (2016), Singapore (2013), South Africa (2014) and the United States (2015). Most of the credit risk sce-narios were applied only to corporate bond expo-sures. Sovereign stress has been added since the European sovereign debt crisis, such as in the case of Luxembourg, Belgium (2013), and Denmark (2014).

• Foreign exchange risk was included in every other exer-cise. Only half of all stress tests included an explicit shock to foreign exchange rates, which can be ex-plained by its relatively small relevance relative to other risk factors. For most exercises, a simple varia-tion of the external value of the domestic currency was assumed, ranging between 15 and 35 percent. Rather severe shocks were applied in the Nether-lands (45 percent depreciation of the euro), Denmark (2007, +/- 40 percent), and South Africa (2008, +/- 50 percent).

• Life underwriting risk was included in fewer than half of the exercises. In most cases, a mortality shock was included with mortality rates exceeding baseline as-sumptions by between 15 and 30 percent. In Spain and Portugal, the effect of lower- than- expected mor-tality rates was also tested. This approach has been developed further by testing higher mortality rates together with increased longevity for annuitants (Guernsey, the Isle of Man, and South Africa [2008]). Pandemics or higher morbidity rates (similar to mor-tality rates, mostly in a range of 15 to 25 percent above the baseline assumptions) were tested in Guernsey, the Isle of Man, Spain, South Africa (2008), and the United States (2010). In a scenario- based motivation of risk factors, higher lapse and sur-render rates were included in five exercises. While earlier stress specifications prescribed a general in-crease in lapse rates (50 percent in Spain and Portu-gal, 30 percent in Guernsey), which could potentially be beneficial to insurers, in the case of Belgium (2013) higher lapse rates (+30 percent) were assumed for those policies for which higher lapses would result in losses for insurers, that is, where the surrender val-ues exceeded the technical provisions.

• Non- life underwriting risks were incorporated mainly via natural catastrophe scenarios. Among the historic sce-narios tested were the Lisbon earthquake of 1755 (Por-tugal) and Hurricane Andrew (United States, 2015); for the latter the claims were assumed to be twice the amount of the historical claims. The FSAP for Bel-gium (2013) included the probable maximum loss ex-pected over a 40-year risk horizon. Other non- life stresses included adverse changes to the claims experi-ence, such as higher claim levels and a higher frequency of large losses (for example, an increase of 10 percent in the cost of claims and a 15 percent higher frequency of claims greater than EUR 30,000 [Spain]), or a worsen-



tion of capital for insurance companies.59 When insurers rely on capital instruments of lower quality, adjustment before or after the stress test— together with a careful interpretation of the results— would be necessary.

The interpretation of stress test results reflects the degree of granularity, the sensitivity of findings to various assump-tions, and the calibration of risk factors under different sce-narios. BU approaches facilitate a more nuanced and detailed capital assessment under stress (including the contribution of individual shocks) due to the direct involvement of firms in the completion of the exercise. Findings are also condi-tional on the mitigating (or aggravating) impact of business and external factors (that is, business strategy and market competition) as well as operational (that is, management be-havior) and structural considerations, for example, a change in policyholder participation or changes in deferred tax as-sets/liabilities (Figure 17.6). If shocks are expected to affect both available and required capital,60 a disaggregated view of the stress test results is desirable. In both FSAPs and national supervisory stress tests, the mitigating (or aggravating) influ-ence of business and external factors tends to be assessed on a qualitative basis only. While management actions and hedging are recognized in many exercises, authorities are

requirement, many jurisdictions have also implemented a minimum capital requirement or a balance- sheet- based minimum solvency margin as a minimum threshold, which, if breached, triggers the strongest supervisory actions, such as business suspension and revocation of licenses.

The definition of solvency in stress tests can deviate from prudential norms and might involve alternative measures that complement the prudential definition of capital ade-quacy. A supervisory stress test might not be an essential component of the national solvency regime but might serve as a sensitivity test only. General accounting- based solvency standards and solvency indicators could be applied, such as the net- premium/ loss- reserve ratio, net-premium-to-capital-andsurplus ratio, and a simple net- asset- value measure (that is, excess assets over liabilities). In an instantaneous stress test, the impact is usually measured by simply comparing preshock solvency with postshock solvency (or an alternative proxy for changes in solvency based on actuarial/accounting indicators).

The assessment of capital adequacy is heavily influenced by the definition of capital resources and their availability under stress within the relevant solvency regime. In addi-tion, in jurisdictions with stringent statutory requirements for current (best) estimates of technical provisions and mar-gin requirements, it might be appropriate to include some reserves in the regulatory eligible capital resources.58 How-ever, including less reliable (nonpermanent) capital instru-ments (such as subordinated debt) and assets (such as intangible assets, deferred tax assets, and deferred acquisi-tion costs) requires careful consideration as to their potential loss absorption capacity given the absence of a clear defini-

58 Current (best) estimate reflects the expected present value of all relevant future outflows that arise in fulfilling insurance obligations, using un-biased, current assumptions.

59 More specifically, ICP 17.11.34 provides only a broad categorization of capital: (1) highest quality capital— permanent capital that is fully avail-able to cover losses of the insurer at all times on a going- concern and wind- up basis, (2) medium quality capital— capital that lacks some of the characteristics of highest quality capital, but which provides a de-gree of loss absorption during ongoing operations and is subordinated to the rights (and reasonable expectations) of policyholders, and (3) low-est quality capital— capital that provides loss absorption in insolvency/ winding- up only.

60 An effect on required capital can usually be assumed in a risk- based sol-vency framework, which is typically the case for market or credit stresses, while for most underwriting effects the effect is rather negligible.

Box 17.7. Insurance Stress Test in the Financial Sector Assessment Program for Belgium

The stress test covered the six largest insurers (representing more than 70 percent of the sector) to determine the capacity of the insurance sector to absorb a combination of single- factor shocks (IMF 2013b). The exercise was conducted by insurers themselves (that is, “ bottom- up”) in collaboration with the IMF Financial Sector Assessment Program (FSAP) team and National Bank of Belgium (NBB) staff based on mid- 2012 prudential data, following the calculation method and guidelines provided by the NBB. The NBB calibrated four market risk factors— interest rates, equity prices, corporate spreads, and sovereign spreads— for a mild and a severe adverse scenario, together with a mass lapse event in the life business and the realization of the largest probable maximum losses on a single ( man- made or natural) cata-strophic tail event.1 The non- life catastrophe event and the life insurance mass lapse were identical for both scenarios. Insurers calculated the overall capital impact by aggregating the individual impact of each risk factor using a correlation matrix, similar to the aggregation approach under the Solvency II.

The amount of own funds available under each scenario was then compared with the solvency capital requirement (SCR) and the mini-mum capital requirement, subject to eligibility conditions.2 However, the main effect of the scenario was its impact on own funds, rather than on SCR. Also the impact of general conditions affecting risk factors, such as the upcoming regulatory reforms, was examined as the sector transitioned from the prevailing Solvency I regime to a more risk- based solvency standard (Solvency II).

1 The tests were carried out using data as of the end of June 2012 for all but one insurer (which used end- of- September 2012 data due to corporate restructuring in the interim period).

2 This is a slight simplification, since the SCR and the minimum capital requirement changed during times of stress.



Source: IMF staff.Note: Boxplots in panel 1 include the asset-weighted and simple average (green dots and yellow diamonds) and the interquartile range (that is, sample results between the 25th and 75th percentiles) (blue boxes). The template for these output charts is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit.

Figure 17.7a Presentation Templates of Outputs (Hypothetical single-period test)

301+ 251–300 201–250 151–200101–150 0–100

All groups and legal entities Only groups Only legal entities (solo basis)

Overall losses in base scenario Overall losses in adverse scenario

0

50

100

150

200

250

300

350

Prestress Baseline AdverseScenario

AdverseScenario

1. Overall Solvency Ratios (In percent)

2. Overall Solvency Ratios (In percent, weighted average)

3. Distribution of Results (by “solvency buckets”) (In percent of total insurance sector assets)

5. Breakdown of Changes in Aggregate Solvency (In percentage points)

4. Individual Risk Impacts (In [currency and unit])

0

50

100

150

200

250

300

Prestress Baseline AdverseScenario

0102030405060708090

100

Prestress Baseline

020406080

100120140160180200

Pres

tress

solv

ency

ratio

Inte

rest

rate

Equi

ty

Spre

ad

Sove

reig

n

Non-

life

insu

ranc

e

Life

insu

ranc

e

Polic

yhol

der

parti

cipa

tion

Defe

rred

taxe

s

Post

stre

ssso

lven

cyra

tio

0

2

4

6

Inte

rest

rate

risk

Equi

tyris

k

Spre

ad risk

Sove

reig

nris

k

Non-

life

insu

ranc

est

ress Life

insu

ranc

est

res

requiring insurers to report the results without recognizing those actions.

A higher aggregation level of available stress test results due to data confidentiality increases the importance of the presentation format.61 In many countries, the insurance

sector can be segmented into life, non- life, and reinsur-ance. Results should be reported for each year of the fore-cast time horizon (with some measure of dispersion, such as the interquartile range, that is, between the 25th and the 75th percentile of the distribution of solvency levels) and certain performance measures. Also the contributions of different risk drivers to the overall solvency results, risk mitigation effects, and recognized diversification effects should be shown. Figure 17.7a shows various graphical pre-sentations of single- period stress test results. In the case of

61 Results related to FSAP stress test are only published after consultation with the country authorities and approval by the IMF Executive Board, subject to the existing confidentiality agreements between national author-ities and the IMF as well as IMF statutes that govern data confidential-ity with authorities.




compared to similar analyses performed on banks. EIOPA refrains from publishing firm- by- firm results of its system- wide stress test, stating that the test “is based on a future regulatory regime and not necessarily indicative of any cur-rent solvency problems (EIOPA 2011a).”63

The FSAP stress test is generally documented in a Techni-cal Note, which is published (together with the FSSA report) after approval by the IMF’s Executive Board in consultation with the national authorities (Appendix Table 17.1.1).64 Com-monly published are solvency ratios before and after stress (depending on the stress test methodology for single stress factors or for the full scenario) together with some informa-tion on their dispersion across the sample, such as the inter-quartile range, minimum, and maximum. The presentation of test results is linked to the description of the structure and predominant business models observed in the sector. In ad-dition, the evolving practice of using multiyear projection horizons requires a thoughtful presentation of the contribu-tion of individual shocks to assess the exact timing and du-ration of the stress. Only very few exercises included the

a multiyear projection, the results are compared against a baseline scenario, which reflects a continuation of the busi-ness and external conditions at the start of the forecast ho-rizon (Figure 17.7b).

The traditional balance- sheet- based output measures can be complemented with additional risk- sensitive indicators of financial performance. For instance, profitability mea-sures, such as changes in underwriting and investment in-come relative to net premiums earned/written, can usefully augment capital assessment. A positive net income in a stress scenario is likely to result in a higher solvency ratio and might counterbalance negative valuation effects for assets and liabilities. However, some accounting measures might be difficult to reconcile with the valuation approach used for the stress testing exercise. Although net income or derived measures, like return on equity or return on assets, are com-mon measures of profitability, there are material inconsis-tencies in the valuation of assets and liabilities.62 These difficulties could be addressed by defining an alternative valuation metric, such as the market- consistent embedded value, which can be calculated as the difference between the market value of assets less the market value of liabilities.

The publication of stress test results by insurance supervi-sors has generally been limited thus far, especially when

62 For instance, under Solvency II, net income is not a regulatory concept, which complicates any effort to align it with profitability indicators de-rived from the statutory balance sheet.

63 EIOPA has maintained this position up to this point, as demonstrated in the publication of the latest stress test (see https://eiopa.europa.eu/pages/ financial- stability- and-crisis-prevention/stress-test-2016.aspx). Also, its predecessor, the Committee of European Insurance and Oc-cupational Pensions Supervisors (CEIOPS 2010), decided against dis-closing firm- specific results.

64 These publications do not include any company- specific information; instead stress test results are published on an aggregated level.

Source: IMF staff.Note: The bubble chart depicts the total balance sheet assets as the diameter of the bubbles; blue bubbles depict life insurers, and green bubbles are non-life insurers. The template for these output charts is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit.

Figure 17.7b Presentation Templates of Outputs (Hypothetical multiple-period test)

Baseline Adverse scenario Interest rate and spread Equity Currency Lapse

0

50

100

150

200

250

300

350

2012 14 15 16 17

1. Solvency Ratios over Time (In percent)

2. Adverse Scenario: Contribution of Stresses to Deviations from Baseline (In percentage points)

3. Cumulative Change in Solvency Ratios and Net Income (In percentage points and [currency and unit], respectively)

–30

–14–12–10–8–6–4–2

02

Solvency Ratio

Net I

ncom

e

–20

–10

0

10

20

2013 14 15 16 1713

–60 –50 –40 –30 –20 –10 0


https://eiopa.europa.eu/pages/

https://eiopa.europa.eu/pages/



the communication of the stress test results is probably just as important as the design and implementation of the stress test itself. Since stress tests should attract attention, and, if applicable, trigger action at the senior decision- making level, it is crucial to present the results in a nontechnical manner.65 Moreover, the sensitivity of the outcome to various assump-tions and key risk drivers should be presented comprehen-sively as a robustness check. If stress tests are completed on a regular basis, standard reporting templates aid comprehen-sion and enhance comparability of stress test results.

Most risk- based solvency regimes incorporate regular stress tests as an indispensable tool to complement the as-sessment of solvency conditions. The frequency of stress tests is determined by the way in which the risk sensitivity of the supervisory framework, the dynamics of the insurance sec-tor, and the changes in external factors influence the validity of the existing design and data collection. During periods of stress or greater market uncertainty (when firms are likely to adjust their asset allocation or hedging becomes more diffi-cult or expensive), frequent sensitivity analyses and/or addi-tional stress tests based on ad hoc data surveys provide a more accurate representation of existing vulnerabilities.

The specification of stress test scenarios should reflect the evolving nature of both risk factors and riskmanagement activities. Even though it might be of interest to track the resilience of the insurance sector under the same scenario each time, there is a risk that firms take measures that would allow them to perform well under a specific scenario. In-stead, stress scenarios should be periodically updated to re-flect the time- varying scope of risk factors and their impact on different insurance activities.

presentation of nonsolvency figures. As an example, the stress test for France included postshock policy yields, and the stress tests for Canada, Denmark (2014), and South Af-rica (2014) analyzed the impact on profitability.

Validation of Results

The prevalence of BU approaches in supervisory stress testing of insurance companies puts a premium on due diligence. While asset- side risks from investments are relatively straight-forward to evaluate even at a very aggregate level, the actuarial assessment of insurance liabilities requires detailed knowledge of different underwriting portfolios and their stochastic prop-erties, which is difficult to achieve in a TD exercise. Thus, most insurance stress tests are completed in collaboration with the respective insurers. Such a BU approach benefits from greater accuracy (due to industry participation) but also risks undermining the consistency of a system- wide exercise if par-ticipating insurers applied different assumptions, rendering the overall stress test results less reliable.

Validation should be performed based on historical expe-rience and comparative analysis/benchmarking. Firm- specific results can be validated against financial soundness indica-tors and prudential measures from supervisory reporting, public disclosure, or other forms of disclosure, such as sur-veys and audits. In addition, peer group analysis would help identify outliers across different firms and business lines. Within each of these analytical approaches, comparisons could be made both point- in- time and based on historic trends, including changes to previous stress test results. Ta-ble 17.2 provides an overview of these dimensions and sug-gests some indicators that could be used for validation.

Communicating of Results

The effectiveness of a stress testing exercise critically depends on the communication and disclosure of the identified vul-nerabilities and associated policy recommendations. In fact,

TABLE 17.2

Overview of Possible Validation ChecksSame Company Peer Group Analysis

Point-in-time Impact of stressDisclosed sensitivities

Impact of stressExposure

Impact of stressDisclosed sensitivities

Deviations from baseline assumptions (premiums, claims, and lapses)

Time series Deviations from historic average RoE/RoA

Deviations from historic dividend payout ratio

Deviations from historic average RoE/RoADeviations from historic average

dividend payout ratioDeviations from baseline assumptions

(premiums, claims, and lapses)

Source: Authors.Note: RoA = return on assets; RoE = return on equity.

65 An early example of utilizing stress test results and directly linking su-pervisory action to these is the “ traffic- light approach” introduced in Denmark in 2001 for life insurers and pension funds. Depending on the outcome of the stress test, the supervisory authority is empowered to limit the risk taking of the firm and require additional reporting (IMF 2007c).



analysis in financial stability assessments as part of bilateral surveillance efforts. However, considerable effort is required to develop common scenarios that reflect the intercon-nectedness between the two sectors (via capital market transactions and intragroup obligations), reflecting the diver-sification effects from their complementary balance sheet structures. These scenarios should comprise (1) different time horizons (aggregate impact of instantaneous/ single- risk factor shocks in insurance stress testing versus average im-pact of scenario- based/ multiperiod sensitivity to multiple risk drivers in bank stress testing), (2) different (and in the extreme, opposite) sensitivities to the same shocks,67 and (3) a broader coverage of risk factors affecting the balance sheet under adverse scenarios.68 There is also a case to be made for close alignment of these stress tests with similar exer-cises completed for pension funds (which was the case in Israel [IMF 2012a]).

The evolution of the insurance industry requires a continu-ous reassessment of stress testing practices. The results of stress tests and the interpretation of associated findings are heavily influenced by the scope and calibration of macro- financial risks, the assessment of vulnerabilities to these risks, as well as both data availability and granularity. These empirical and technical challenges raise the following challenges:

• Risk factors are bound to change over time, which can affect the robustness of stress test results. The calibration of risk factors in stress tests is premised on a compre-hensive assessment of the impact from adverse changes in general conditions and trends in the in-surance industry and the broader financial system, the interconnections between insurers and other fi-nancial institutions (with a focus on NTNI activities in insurance groups), and general capital market conditions. Understanding the differences in busi-ness models and behavioral characteristics under stress is fundamental to the qualified assessment of their influence on evolving risk transmission chan-nels affecting the insurance sector.

• The robustness of valuation methodologies may be un-dermined by the severity of stress they are designed to measure. Systemic risks affecting financial stability generally arise from uncertainty, that is, rare and nonrecurring events (with a weakly defined or un-known underlying distribution) rather than repeated realizations of predictable outcomes (with a known distribution). This reality might require a wide range of assessment methods (including qualitative ap-proaches) and cross- validation due to the limited usefulness of certain (quantitative) measures and ac-tuarial valuation models based on robust statistics

The publication of stress test results should be balanced against the consequences of negative market reactions and requires suitable backstops to enhance credibility of follow- up actions. Findings from a stress testing exercise, especially if completed with the participation of firms, tend to consti-tute privileged information, which can materially influence investor behavior and market prices. Thus, the scope of disclo-sure warrants a commensurate communication plan together with an effective recovery and resolution regime, including the option of a readily available fund to protect policyhold-ers of companies that show a capital shortfall under stress. Experience from past banking sector stress tests suggests that comprehensive and transparent disclosure with a public commitment to potential recapitalization results in positive market reactions.66

4. DISCUSSION AND CONCLUSION System- wide stress tests have become increasingly important in the wake of the global financial crisis for micro- and mac-roprudential surveillance of the insurance sector. While in-surance companies are generally less systemically relevant than banks, the interlinkages between insurers, banks, and other financial institutions may increase in the future through products, markets, and organizational arrangements. Na-tional supervisory authorities are revisiting existing stress testing practices with a view toward enhancing their effec-tiveness and usefulness for forward- looking capital assess-ments. Nevertheless, most approaches remain focused on the solvency of individual firms rather than the system- wide ro-bustness to the joint impact of risk factors during times of stress arising from (1) the growing complexity of the inter-connectedness among insurance companies and with other financial institutions, and (2) the extent to which such inter-linkages cause potential spillover and contagion effects. As more jurisdictions move toward market- consistent solvency regimes, regular stress testing can inform a thematic review of key vulnerabilities to risk factors specified by the regula-tory framework.

A more integrated stress testing approach would ideally be based on a common framework for both banking and insurance sectors, or at least draw on the same assumptions (and calibration of risk factors). The closer coordination be-tween banking and insurance stress testing in recent FSAPs, such as in the case of Belgium, Canada, and the United States, underscores the critical role of insurance- sector

66 The publication of the results from the first comprehensive banking stress test in the United States, the Supervisory Capital Assessment Pro-gram in 2009, was well received by investors, largely because the US Federal Reserve’s commitment to recapitalize failing banks was consid-ered sufficiently credible. In contrast, a similar exercise carried out by the European authorities (Committee of European Banking Supervi-sors) in the same year failed to allay investor concerns about existing vulnerabilities in the banking sector, as some relevant risks had been excluded from the exercise that were amplified by the absence of credi-ble backstop measures (Ong and Pazarbasioglu 2014).

67 For instance, a positive shock to interest rates tends to generate higher levels of solvency among insurers, especially long- term underwriters, whereas the opposite holds true for banks.

68 The insurance sector stress tests would include many liability side risks in addition to the asset side risks that affect both insurers and banks alike.



As insurance stress testing continues to evolve, several is-sues need to be considered:

• Most stresses impact both assets and liabilities and should be assessed using a total balance sheet approach. While the modeling of interest rate shocks affecting both sides of the balance sheet has become common practice and features very prominently in stress tests, most notably in the context of market- consistent valuation frameworks, other important macro- financial transmission channels affecting reserve adequacy of both life and non- life insurers should not be over-looked. Claim patterns are closely linked to changes in inflation, and premium risk is influenced by mon-etary conditions. Similarly, in some lines of business such as credit insurance, claims increase significantly during recessions. Depending on the circumstances, the appropriate calibration of shocks affecting liabil-ities often requires weighing the benefits of pre-scribed parameters (including premium and claim developments) against the plausibility (and suffi-cient rigor) of firms’ own assumptions for modeling these effects.

• The aggregate risk impact should not include diversifi-cation benefits among risk factors except where econom-ically plausible. If risk factors are not fully correlated, it is reasonable to account for their dependence structure and combine stress testing parameters so that the individual contribution of each risk to the capital impact is lower than the appropriate percen-tile for that risk in isolation. However, combining multiple risk factors with diversification effects un-der different scenarios tends to complicate a reliable capital assessment under stress. The frequent use of correlation to determine the joint impact of risk fac-tors could be mathematically inconsistent and lead to an underestimation of potential losses.72 The sim-ple aggregation of risk factor impacts would preserve the stochastic assumptions of each risk factor.

• Different combinations of risk factors under alternative scenarios introduce an element of flexibility. The single- risk- factor capital impacts can be used to assess how different combinations of risk factors affect estimates of system- wide capital adequacy under stress.73 Ex-tending the analysis beyond single- factor scenarios may also include the capital assessment of insurers to shocks at different magnitudes in changes of risk factors (that is, subject to varying levels of statistical confidence).

(which tend to rely on the convergence of prices and parameters with long- term expectations).

• The interpretation of macro- financial shocks and their impact on capital adequacy involves a trade- off be-tween accuracy and timeliness. The historical sensitiv-ity of sample firms to macro- financial shocks is essential to assessing the combined impact of se-lected risk factors over a predefined forecast horizon of stress. While reliance on past experiences en-hances confidence in the predictability of how shocks impact capital ex ante, it may also make it difficult to interpret signals in time and provide early warnings without hindsight bias. Conversely, effec-tive early warnings would entail greater uncertainty due to a weaker empirical robustness. Any early warning gains greater accuracy only as the realiza-tion of the identified risk becomes more probable, which limits the flexibility in designing and imple-menting effective policy measures.69

There is a clear trend toward a more precise and consis-tent assessment of vulnerabilities in stress testing models due to greater convergence of regulatory standards and supervi-sory practices. The current work of the IAIS on developing a global solvency regime as part of the Common Framework for the Supervision of Internationally Active Insurance Groups will further influence the methodological framework of scenario- based capital assessment of insurance companies,70 which includes the development and field testing of a risk- based, global insurance capital standard for internationally active in-surance groups (IAIGs) (IAIS 2013e, 2018a, 2018b). This effort includes a cross- country stress testing component to-gether with a BU exercise with participating IAIGs, which promotes the convergence of key concepts, such as economic valuation, as well as the categorization and calibration of risk factors for capital purposes. Also, the introduction of Sol-vency II in the European Union has influenced the design of stress tests in EU Member States (and Solvency II- equivalent jurisdictions)71 over recent years, and is paving the way for a more comprehensive assessment of risk factors (including more shocks and a higher level of granularity) and greater convergence in both taxonomy and methodology.

69 Borio, Drehmann, and Tsatsaronis 2012 state categorically that “stress tests failed spectacularly when they were needed most: none of them helped to detect the vulnerabilities in the financial system ahead of the recent financial crisis.” They concede, however, that stress tests may have a role as crisis management and resolution tools.

70 The Common Framework for the Supervision of Internationally Active Insurance Groups is a set of international supervisory requirements focusing on the effective group- wide supervision of IAIGs, which builds and expands upon the high- level standards and guidance currently set out in the ICPs at a legal entity basis and group-wide level (see https://www .ia isweb.org/page/ supervisory- material/ common-framework/f ile /76033/comframe-frequently -asked-questions).

71 Bermuda and Switzerland are the only two fully equivalent jurisdictions (https://eiopa.europa.eu/ external-relations/equivalence/overview-of -equivalence-decisions).

72 Given that large shocks are transmitted across entities differently than small shocks, the use of nonlinear dependence can deliver more reliable insights about the joint tail risks that arise in extreme loss scenarios (Jobst 2013b).

73 The nonlinearity in the price changes of certain products (such as finan-cial derivatives, embedded options, and nonproportional reinsurance contracts) needs to be taken into consideration for a proper assessment of the impact.


https://www.iaisweb.org/page/


https://eiopa.europa.eu/


insurance entities within groups and conglomerates. However, in some cases, the differential impact of reporting levels can be very material for the capital and liquidity positions of legal entities.

• Secondary impact of financial distress can be material. For instance, the deteriorating solvency position raises the cost of capital and constrains underwriting (especially if a rating downgrade occurs), which lim-its the ability of insurers to generate sufficient pre-mium income and overall profit. A detailed analysis of any secondary impact should be linked to contin-gency as well as recovery and resolution planning.

• The framework of supervisory stress testing should be de-signed with a view to avoiding distortive effects on the behavior of insurers (such as uneconomic changes of the asset allocation or product design). For instance, if stress tests are applied only to the asset side, insurers may reduce the duration of assets; this would im-prove the solvency position after a positive interest rate shock but increase maturity mismatches. If an undue cut of policyholder dividends is recognized in stress tests, insurers may provide more participating products but could end up paying significant divi-dends to protect their reputations even in stress situ-ations. In the same vein, a variation of stress test scenarios over time will likely reduce the risk of in-surance companies trying “to game” the stress test.

• While the constant evolution of risk analytics is likely to create bias toward enhancing stress testing models, qualitative elements cannot be ignored. The dynamics of business strategies, including but not limited to evolving underwriting practices, changes in business models, and risk transfer innovations, require a peri-odic reassessment of the relevance of risk factors and comprehensiveness of the chosen stress testing ap-proach. Stress tests invariably mirror the evolution of risk management; however, more sophisticated meth-ods require (more) granular data—which will never be sufficient for reliably modeling tail risk. Thus, ex-pert judgment will remain a highly crucial element of stress tests. This also places greater focus on more qualitative analysis, such as the reputational risk of individual firms, the competitive environment, and existing risk controls that influence the gross impact of risks.

• Additional performance indicators (outside the initial scope of the stress test parameters) can provide a more comprehensive perspective on the full impact of different stress scenarios. Accounting measures (for example, net income and other profitability indicators) affect the decision to pay out dividends to shareholders or bo-nuses to policyholders, which can influence the assess-ment of adequate solvency buffers under stress. Also incorporating liquidity measures could provide useful insights, especially when investment assets become more illiquid (for example, due to lower market li-quidity in an adverse scenario) or when the cash- flow demands of a particularly severe claims shock (for ex-ample, a mass lapse event or a catastrophe) might af-fect the counterbalancing capacity of liquid assets.

• The extension of single- period shocks to multiperiod sce-narios helps identify medium- to long- term vulnerabili-ties. Extending the stress test horizon and applying multiyear scenarios illustrates emerging vulnerabili-ties from a gradually eroding solvency position over time, which would inform suitable remedial actions and recovery plans. Moreover, even the market- consistent valuation of liabilities might not fully cap-ture the uncertainty of future cash outflows related to insurance claims (for example, most liabilities for asbestos- related claims were only recognized decades after the contract was issued), which could be ad-dressed in multiperiod scenarios.

• When performed in parallel, banking and insurance stress tests can be reconciled by means of using the same target variables. A common metric of risk factors and shocks allows for an integrated analysis at a system- wide level but also at the level of a conglomerate. While the impact of a given scenario defined by changes in economic activity, asset prices, and inter-est rates is likely to differ between insurance and banking activities, supplementary sensitivity analy-ses for the less affected sector can usually provide ad-ditional insights.

• The impact of scenarios on intragroup transactions should be assessed by comparing stress testing results for different reporting levels. Stress tests based on consoli-dated reporting do not include sufficiently granular data to identify vulnerabilities from intragroup transactions or transactions between banking and


Appendix 17.1.Tables


Overview of Insurance Stress Tests in FSAPsFSAP Mission Dates Publication

Japan June and October 2002, March 2003 FSSA: September 2003 (IMF 2003)Singapore November 2002, July/August and September 2003 FSSA: April 2004 (IMF 2004c)Netherlands October/November 2003, March 2004 FSSA: September 2004 (IMF 2004a)France February 2004, May 2004 FSSA: November 2004 (IMF 2004b)

TN: June 2005 (IMF 2005)Belgium December 2004, March 2005 FSSA: February 2006 (IMF 2006a)Spain June/July and October/November 2005 FSSA: June 2006 (IMF 2006b)

TN: June 2006 (IMF 2006c)Denmark November 2005, May 2006 FSSA: October 2006 (IMF 2006d)

TN: March 2007 (IMF 2007b)Mexico February/March 2006 FSSA: October 2006 (IMF 2006e)

TN: May 2007 (IMF 2007f)Portugal December 2005, May 2006 FSSA: October 2006 (IMF 2006f)

TN: January 2007 (IMF 2007a)Switzerland November 2006 FSSA: June 2007 (IMF 2007d)

TN: June 2007 (IMF 2007e)Bermuda June 2007 FSSA: October 2008 (IMF 2008b)South Africa May 2008 FSSA: October 2008 (IMF 2008a)Isle of Man September 2008 FSSA: September 2009 (IMF 2009a)

TN: September 2009 (IMF 2009b)United States October/November 2009, February/March 2010 FSSA: July 2010 (IMF 2010a)

TN: July 2010 (IMF 2010b)Guernsey March 2010 FSSA: January 2011 (IMF 2011a)

TN: January 2011 (IMF 2011b)Luxembourg November 2010 FSSA: June 2011 (IMF 2011e)Israel November 2011 FSSA: April 2012 (IMF 2012a)

TN: April 2012 (IMF 2012c)Japan November/December 2011, March 2012 FSSA: August 2012 (IMF 2012d)France January and June 2012 FSSA: December 2012 (IMF 2012e)Belgium November 2012, January 2013 FSSA: May 2013 (IMF 2013a)

TN: May 2013 (IMF 2013b)Singapore May and July/August 2013 FSSA: November 2013 (IMF 2013d)Canada June and September 2013 FSSA: February 2014 (IMF 2014a)

TN: March 2014 (IMF 2014b)Denmark March and June/July 2014 FSSA: December 2014 (IMF 2014c)

TN: December 2015 (IMF 2014d)South Africa April, May, and June 2014 FSSA: December 2014 (IMF 2014e)

TN: March 2015 (IMF 2015a)United States October/November 2014, February/March 2015 FSSA: July 2015 (IMF 2015b)

TN: July 2015 (IMF 2015c)Norway February/March 2015 FSSA: September 2015 (IMF 2015d)

TN: September 2015 (IMF 2015e)Germany November 2015, February/March 2016 FSSA: June 2016 (IMF 2016a)

TN: June 2016 (IMF 2016b)

Sources: IMF staff; and authors.Note: FSAP = Financial Sector Assessment Program; FSSA = Financial System Stability Assessment; TN = Technical Note.




Overview of National Supervisory Stress Testing ApproachesReference

Austria Austria: Financial Sector Assessment Program Update, Technical Note—Factual Update and Analysis of the IAIS Insurance Core Principles (IMF 2008c)

Bermuda Stress/Scenario Analysis (Class 4, Class 3B and Insurance Groups) and Stress/Scenario Analysis (Class 3A) (BMA 2013a, 2013b; Appendix 17.2)

Canada Stress Testing Guideline (OSFI 2009)Czech Republic Models for Stress Testing in the Insurance Sector (Komárková and Gronychová 2012)Denmark Financial Sector Assessment Program—Detailed Assessment of Observance of the Insurance Core

Principles (IMF 2007c)European Union Specifications for the 2011 EU-Wide Stress Test in the Insurance Sector (EIOPA 2011b)Germany Conducting of Stress Test (BaFin 2004)Guernsey Stress Testing of the Guernsey Insurance Sector (Guernsey Financial Services Commission 2011)Japan Supervisory Guidance for Insurers (JFSA 2013)Singapore ERM Notice (MAS 2011), Stress Testing on Financial Condition of Life Direct Insurer (MAS 2013)Switzerland White Paper of the Swiss Solvency Test (Swiss Federal Office of Private Insurance 2004)United Kingdom Stress and Scenario Testing (FSA 2008)United States Own Risk and Solvency Assessment (ORSA) (NAIC 2013)

Sources: IMF staff; and authors.


Andreas A



479


Stress Testing Matrix (STeM): Insurance Stress Testing Approaches in IMF FSAPs (as of December 2016)1

Country Japan Singapore Netherlands

Year (FSSA publication) 2003 2003 2004

1. Scope Approach BU BU BU Coverage 10 life 10 insurers n/a Relevance of the coverage 86 percent (life, based on assets) 77 percent (life), 45 percent (non-life)2 54 percent (based on assets) Reporting basis n/a n/a n/a Data sources Public n/a Prudential

2. Valuation Basis Valuation of assets and liabilities n/a n/a n/a

3. Scenario DesignMacro-financial linkage/

transmission channel(s)Single-factor shocks Scenario analysis Single-factor shocks

Risk horizon Single period Single period Single period

4. Risk Factors Assets Credit risk 1.5 percent credit loss

on loan bookUp to -7 percent in corporate bond prices

(Singapore and other SE Asian markets); up to -3 percent rest of the world

Up to +50 percent credit spread increase for investment grade; up to +60 percent for

speculative grade; +25 percent implied volatility Equity risk -20 percent Up to -20 percent (Singapore and other

SE Asian markets); -5 percent rest of the world

Up to -40 percent for developed countries; up to -50 percent for developing countries and private equity; +25 percent implied volatility

FX risk — Between -3 percent and +3 percent against various currencies

Up to 45 percent depreciation of the euro; +25 percent implied volatility

Real estate risk — Up to -20 percent (Singapore commercial); -5 percent rest of the world

Up to -20 percent

Interest rate risk +100 bps parallel shift -60 bps in short-term rates (with unchanged long-term rates); +150 bps in short-term rates

(long-term rates +50 bps)

+/-100 bps parallel shift; +/-200 bps parallel shift; +25 percent implied volatility

(continued)



480

APPENDIX TABLE 17.1.3 (continued )


Country Japan Singapore Netherlands

Year (FSSA publication) 2003 2003 2004 Liabilities Life underwriting Mortality/morbidity/longevity — — — Lapse/surrender rates — — — Non-life underwriting Natural catastrophe (based on PML) — — — Other non-life underwriting shocks — — — Other risks — Receivables (outstanding premiums and

agents’ balances): up to -20 percent; loans and other receivables up to -10 percent

50 percent increase in worst technical result in last five years; 50 percent increase in maximum

cost in last five years; up to -45 percent in commodity prices; +25 percent implied

volatility in commodities Risk aggregation/diversification effects Simple summation Simple summation and aggregation with

correlation of 0.5 between all shocks

5. Regulatory Capital StandardsLoss measured as percentage

of shareholder equityMinimum solvency margin requirements Solvency I

6. Presentation of Results Business segment/peer group

breakdown— — —

Dispersion measures Distribution of losses as a percentage of shareholder

equity

— —

Contribution of individual shocks Impact of individual shocks on shareholder equity

— Impact of individual shocks on solvency ratio

Other — — —

(continued)


Andreas A



481



Country France Belgium Spain


1. Scope Approach BU BU BU Coverage 26 life, 52 non-life 2 life and 4 bancassurance groups 27 insurers Relevance of the coverage 79 percent (life), 75 percent (non-life)3 76 percent 62 percent (life), 50 percent (non-life) Reporting basis Solo/consolidated Solo/consolidated Solo/consolidated Data sources n/a Prudential Prudential

2. Valuation Basis Valuation of assets and liabilities n/a n/a Market-consistent

3. Scenario Design Macro-financial linkage/transmission

channel(s)Single-factor shocks Single-factor shocks Single-factor shocks


4. Risk Factors Assets Credit risk — Credit spreads +50 bps Realization of implied PD of Basel II

risk-weights; alternatively credit spread shock

Equity risk -30 percent -30 percent n/a FX risk — n/a Real estate risk -30 percent -20 percent -17 percent Interest rate risk +/-100 bps and +300 bps parallel shift;

steepening of yield curve; flattening and upward shift of yield curve

+200 bps parallel shift +/-200 bps parallel shift; steepening of the yield curve; flattening of the yield curve

(continued)



482

(continued)



Country France Belgium Spain

Year (FSSA publication) 2005 2006 2006 Liabilities Life underwriting Mortality/morbidity/longevity — — Mortality (+/-15 percent), morbidity

(+/-15 percent) Lapse/surrender rates — — +50 percent Non-life underwriting Natural catastrophe (based on PML) Doubling the claims of a storm event

in 1999Yes —

Other non-life underwriting shocks — — +10 percent average cost of claims; +15 percent higher frequency of claims

> 30,000 euros Other risks — 50 percent worsening technical result;

50 percent increase operating costs—

Risk aggregation/diversification effects Various simple summations of individual single-factor shocks

— —

5. Regulatory Capital StandardsSolvency I Solvency I Solvency I

6. Presentation of Results Business segment/peer group breakdown Split into life/non-life — Split into life/non-life/mixed institutions

Dispersion measures Min/max impact on solvency ratio; min/max solvency ratios after stress;

min/max policy yields after stress

Min/max impact on solvency and operating profit Min/max impact on insurers’ capital, standard deviation

Contribution of individual shocks Impact of various scenarios that present different combinations of

individual single-factor shocks

Impact of individual shocks Impact of individual shocks

Other Number of companies with solvency ratios after stress below 100 percent and recapitalization need as percent

of liabilities for these companies; number of companies with a “policy yield shortfall,” and yield shortfall to

liabilities

Impact of natural catastrophe shock before and after reinsurance

—


Andreas A



483



Country Denmark Mexico Portugal


1. Scope Approach BU BU TD BU Coverage Five largest life n/a Four non-life Four life, two non-life,

three composite Relevance of the coverage 50 percent3 n/a 48 percent2 78 percent (life), 64 percent

(non-life)2

Reporting basis n/a n/a Solo/consolidated Solo/consolidated Data sources n/a n/a n/a n/a


3. Scenario Design Macro-financial linkage/transmission channel(s) Single-factor shocks Single-factor shocks Single-factor shocks Risk horizon Single period Single period Single period

4. Risk Factors Assets Credit risk — — — Realization of implied PD

of Basel II risk-weights Equity risk -30 percent — — +/-35 percent FX risk +/-40 percent — — +/-15 percent Real estate risk -30 percent — — +/-5 percent Interest rate risk +250 bps parallel shift;

-100 bps parallel shift-190 bps in domestic interest rates;

-114 bps in foreign interest rates— +/-94 bps parallel shift

of entire yield curve Liabilities Life underwriting Mortality/morbidity/longevity — — — Mortality (+/-15 percent)

(continued)



484



Country Denmark Mexico Portugal

Year (FSSA publication) 2007 2007 2007 Lapse/surrender rates — — — +/-50 percent Non-life underwriting Natural catastrophe (based on PML) — — Earthquake with a

probability of 1/250 years

—

Other non-life underwriting shocks — — — — Other risks — Premium shock: zero nominal premium

growth; loss rate increase: life (10 percent), accidents and health (13 percent), P&C (20 percent), auto (5 percent), and catastrophe

(10 percent)

— —

Risk aggregation/diversification effects — Simple summation (premium shock + interest rate shock; premium shock + loss rates;

premium shock + interest rate shock + loss rates)

Combination with BU results (market and

life stress) via correlation matrix

(QIS2 of Solvency II)

—

5. Regulatory Capital StandardsSolvency I n/a Solvency II SCR

6. Presentation of Results Business segment/peer group breakdown — — — Dispersion measures — Distributions of solvency ratios before and

after shockAnonymized company-by-company data (for the

catastrophe module: absolute gross and net losses, gross and net losses to capital surplus; for the

combined BU and TD impact: reduction in capital surplus)

Contribution of individual shocks Impact of individual shocks on available capital, required capital

and solvency ratio

Impact of individual shocks on solvency ratio

Impact of individual shocks on free surplus

Other — Capital shortfall as percent of market solvency requirement

—

(continued)


Andreas A



485



Country Switzerland Bermuda South Africa


1. Scope Approach BU BU TD BU Coverage Nine life, 12 non-life, nine health 10 large commercial and long-terms Four largest life Relevance of the coverage n/a n/a 55 percent of life2

Reporting basis Solo/consolidated Solo Solo Data sources Prudential Prudential Prudential/public

2. Valuation Basis Valuation of assets and liabilities Market-consistent Statutory accounting Statutory accounting




4. Risk Factors Assets Credit risk GBP credit spread (+50 bps) Yes Credit spreads (+50 percent, except

governments) Equity risk -30 percent Yes -35 percent (and +100 percent implied

volatility) FX risk +20 percent (CHF against EUR, GBP

and USD)+/-50 percent (and +100 percent implied

volatility) Real estate risk -20 percent — -50 percent Interest rate risk 25 bps (short-term)/75 bps (long-

term) lower than the lowest interest rates in the last economic cycle

Yes +600 bps and -400 bps parallel shift (and +100 percent implied volatility)

(continued)



486



Country Switzerland Bermuda South Africa

Year (FSSA publication) 2007 2008 2008 Liabilities Life underwriting Mortality/morbidity/longevity n/a Pandemic events Mortality (+30 percent)

Morbidity (+30 percent) Longevity of annuitants (+30 percent)

Lapse/surrender rates — — — — Non-life underwriting Natural catastrophe (based on PML) n/a Yes — — Other non-life underwriting shocks n/a — — — Other risks — — — — Risk aggregation/diversification effects Correlation matrix of Swiss Solvency

TestSimple summation and correlation

(diversification effect)— —

5. Regulatory Capital StandardsSwiss Solvency Test Change in capital and surplus,

minimum regulatory premium ratio, and minimum regulatory loss reserve

ratio

Total capital divided by total assets

6. Presentation of Results Business segment/peer group breakdown Split into life/non-life/health — — Dispersion measures Boxplots with single data points per

anonymized company (change in risk-bearing capital)

— Min./max impact on solvency ratio of individual shocks (in percentage points)

Contribution of individual shocks Impact of individual shocks on solvency ratio (change in risk-bearing capital)

— Impact of individual shocks on solvency ratio

Other — — —

(continued)


Andreas A



487



Country Isle of Man United States Guernsey


1. Scope Approach BU BU BU Coverage Six largest life 30 largest life All except pure captives Relevance of the coverage 82 percent3 68 percent2 n/a Reporting basis Solo/consolidated n/a Solo/consolidated Data sources Prudential n/a Prudential

2. Valuation Basis Valuation of assets and liabilities Statutory accounting Statutory accounting Statutory accounting


channel(s)Single-factor shocks Scenario analysis Scenario analysis + single-factor shocks

Risk horizon Single period Five years Single period

4. Risk Factors Assets Credit risk Downgrade by two–four notches n/a Downgraded by two–four notches Equity risk -35 percent n/a -35 percent FX risk +/-20 percent (GBP) — Up to +/- 30 percent (GBP) Real estate risk — n/a -20 percent Interest rate risk +/-200 bps parallel shift of entire yield

curven/a Up to +/- 300 bps parallel shift of entire

yield curve

(continued)



488



Country Isle of Man United States Guernsey

Year (FSSA publication) 2009 2010 2011 Liabilities Life underwriting Mortality/morbidity/longevity Mortality (+25 percent), morbidity

(+25 percent), longevity of annuitants (+25 percent)

Pandemic (equivalent to 100 percent RBC) Mortality (+25 percent), morbidity (+25 per-cent), longevity of annuitants (+25 percent)

Lapse/surrender rates — — +30 percent Non-life underwriting Natural catastrophe (based on PML) — — n/a Other non-life underwriting shocks — — n/a Other risks — — — Risk aggregation/diversification effects — Simple summation Simple summation

5. Regulatory Capital StandardsRMM (required minimum margin) Risk-based capital (RBC) Excess assets over liabilities (net asset value)

6. Presentation of Results Business segment/peer group breakdown — Qualitative description of effect on

variable annuity providers—

Dispersion measures Min./max. change in solvency ratio (in percentage points)

— Min./max. impact on excess of assets over liabilities

Contribution of individual shocks Impact of individual shocks on solvency ratio

— Impact of individual shocks on excess of assets over liabilities

Other — Number of companies below 300 percent RBC

Number of companies with negative net asset value

(continued)


Andreas A



489



Country Luxembourg France Israel


1. Scope Approach BU and TD BU BU Coverage All insurers 12 life and unknown number of non-life All insurers Relevance of the coverage 100 percent 70 percent (life) 100 percent Reporting basis n/a Solo n/a Data sources n/a n/a n/a





4. Risk Factors Assets Credit risk Sovereign distress; for non-life

companies: failure of largest depository bank

n/a Credit spread (between +50 and +200 bps)

Equity risk -25 percent Yes Up to -30 percent FX risk — — +/-20 percent Real estate risk -15 percent Yes — Interest rate risk +25 percent along yield curve Yes +/-20 percent of the risk-free rate Liabilities Life underwriting Mortality/morbidity/longevity — Yes — Lapse/surrender rates — Yes —

(continued)



490



Country Luxembourg France Israel

Year (FSSA publication) 2011 2012 2012 Non-life underwriting Natural catastrophe (based on

PML)Deterioration in the claims situation n/a n/a

Other non-life underwriting shocks

n/a n/a n/a

Other risks — — — Risk aggregation/diversification

effectsCombined equity and interest rate

shock via (1) simple summation, and (2) QIS5 correlations

n/a n/a

5. Regulatory Capital StandardsSolvency I Solvency I n/a


breakdownSplit into life/non-life; split into fully

reinsured and not fully reinsured companies

Split into life/non-life Split by product lines (participating policies, provident funds, new pension funds)

Dispersion measures Number of companies in poststress solvency buckets, including the

average preshock solvency ratio for each bucket

Min/max impact on aggregate solvency ratio of individual shocks (in percentage

points)

Anonymized company-by-company data (change in capital surplus in percentage points)

Contribution of individual shocks Impact of individual shocks on solvency ratio

Impact of individual shocks and shock absorption mechanisms

Impact of individual shocks on value of long-term savings

Other — — Impact on value of long-term savings

(continued)


Andreas A



491



Country Japan Belgium Singapore


1. Scope Approach BU BU BU Coverage Four life, five non-life Six largest life Four largest life Relevance of the coverage 43 percent (life), 82 percent

(non-life)70 percent 80 percent

Reporting basis Solo Solo/consolidated Solo Data sources Prudential/public Prudential Prudential/public

2. Valuation Basis Valuation of assets and liabilities Statutory accounting with some

adjustments of unrealized gain/loss and economic valuation for

interest rate sensitivity

Statutory accounting, quasi-Solvency II (QIS5), and market-consistent

Market-consistent


channel(s)Scenario analysis + single-factor

shocksSingle-factor shocks Scenario analysis + single-factor shocks

Risk horizon Two years Single period Three years

4. Risk Factors Assets Credit risk 3 percent (1.5 percent for life) credit

loss on loan book +80 percent of reinsurers’ failures

Credit spread (between +30 bps and +1,260 bps)

Credit spread (between +25 bps and +300 bps)

Equity risk -20 percent Between -16 and -23.7 percent Between -15 and -30 percent FX risk — — Between +5 and -35 percent (in first year) Real estate risk — — — Interest rate risk +100 bps parallel shift Between -61 and - 82 bps parallel shift Up to +150 bps

(continued)



492



Country Japan Belgium Singapore

Year (FSSA publication) 2012 2013 2013 Liabilities Life underwriting Mortality/morbidity/longevity Yes (pandemic increasing mortality

rate of 0.13 percentage points)— —

Lapse/surrender rates — +30 percent4 — Non-life underwriting Natural catastrophe (based on PML) — Largest PML on a single catastrophe event

(1/40 year event)—

Other non-life underwriting shocks Moderate reinsurance failure — — Other risks — — — Risk aggregation/diversification effects Simple summation Simple summation and correlation matrix

similar to that of Solvency IISimple summation

5. Regulatory Capital StandardsSolvency Margin Ratio Quasi-Solvency II SCR/MCR Singapore Risk-Based Capital

6. Presentation of Results Business segment/peer group breakdown Split into life/non-life — — Dispersion measures — — — Contribution of individual shocks Separate presentation of the effects

of the pandemic shock and the reinsurance failure

Capital impact of individual shocks —

Other Composition of change in solvency margin (net realized gains/losses on securities, unrealized gains/

losses on land, contingency reserve, price fluctuation reserve,

and others)

Prestress composition of solvency capital requirements; poststress solvency ratios

in percent of prestress level

Number of companies below 100 percent statutory minimum capital requirements, both for one-year and three-year horizon

(continued)


Andreas A



493



Country Canada Denmark South Africa


1. Scope Approach BU BU BU Coverage Three largest life Four life, four non-life Four life, four non-life Relevance of the coverage 60 percent2 50 percent (life), 65 percent (non-life)2 70 percent (life, based on assets), 50 percent

(non-life1) Reporting basis Consolidated Consolidated Solo Data sources Prudential Prudential Prudential

2. Valuation Basis Valuation of assets and liabilities Market-consistent Quasi-Solvency II Statutory accounting


channel(s)Scenario analysis + single-factor

shocksScenario analysis + single-factor shocks Scenario analysis + single-factor shocks

Risk horizon Five years Five years Five years

4. Risk Factors Assets Credit risk Credit spread (increase to a level three

to four times higher than end-2012)Credit spread (domestic sovereign -3 bps, between -22 bps and +399 bps for other sovereigns; between +39 bps and +1,178

bps for corporate bonds); sensitivity analysis: +500 bps spread increase for

covered bonds

Credit spread (domestic sovereign +262 bps, between +148 bps and +321 bps for corporate

bonds), default of largest banking counterparty

Equity risk Between -21 percent and -49 percent in first year

-27 percent -50 percent for ordinary shares, up to -30 percent for preference shares

FX risk Between +10 percent and +20 percent appreciation of USD

No change in DKK/EUR, appreciation of EUR against other Nordic currencies, GBP,

and USD

30 percent depreciation of ZAR

Real estate risk Between -34 percent and -54 percent over three years

-11 percent -30 percent

Interest rate risk Sharp decline in first year, followed by gradual increase and steepening in

subsequent years

Up to -102 bps in first year (DKK); sensitivity analysis: sharp increase

+50 percent parallel shift; sensitivity analysis: -35 percent parallel shift

(continued)



494



Country Canada Denmark South Africa

Year (FSSA publication) 2014 2014 2014 Liabilities Life underwriting Mortality/morbidity/longevity — — — Lapse/surrender rates +20 percent in first year Yes — Non-life underwriting Natural catastrophe (based on PML) — Historic event (windstorm Anatole, 1999),

1/100 PMLLargest PML on a single catastrophe event (1/100 year event), default of reinsurer with

45 percent LGD Other non-life underwriting shocks — — — Other risks — — — Risk aggregation/diversification effects Simple summation Simple summation Simple summation

5. Regulatory Capital StandardsMCCSR (minimum continuing capital

and surplus requirement)The higher of 1) Solvency I requirements, and 2) QIS5-like Solvency II requirements

Statutory solvency regime (capital adequacy requirement)


breakdown— Split into life/non-life Split into life/non-life

Dispersion measures — — Interquartile distribution for effect on solvency ratio and return on equity

Contribution of individual shocks Capital impact of individual shocks (for each year of the five-year

projection horizon)

Capital impact of individual shocks Capital impact of individual shocks

Other Impact on net income (for each year of the five-year projection horizon)

Impact on net income Impact on return on equity

(continued)


Andreas A



495



Country United States Norway Germany


1. Scope Approach TD BU TD Coverage 20 life, 16 non-life, five health, two credit

insurersThree life, three non-life 75 life

Relevance of the coverage 40 percent2 80 percent (life, based on assets), 60 percent (non-life2)

93 percent (life, based on assets)

Reporting basis Consolidated Solo Solo Data sources Public Prudential Prudential

2. Valuation Basis Valuation of assets and liabilities US GAAP; statutory accounting Quasi-Solvency II Solvency II


channel(s)Scenario analysis + single-factor shocks Scenario analysis + single-factor shocks Scenario analysis + single-factor

shocks Risk horizon Single period; five years for low-for-long Single period Single period

4. Risk Factors Assets Credit risk Credit spread (US sovereign +58 bps; US municipal

bonds between +66 bps and +429 bps; other sovereign bonds between +57 bps and +1,661 bps; corporate bonds between +248 bps and +1,878 bps; market value loss in CMBS, RMBS

and other ABS between -28.7 percent and 51.3 percent; mortgage loans -12.2 percent)

Credit spread (between +70 bps and +700 bps), default of largest banking

counterparty

Market value of corporate bonds between -4.5 percent and -37.5 percent; credit spread

between +25 bps and +100 bps

Equity risk -28.9 percent Between -26 percent and -55 percent Between -22 percent and -49 percent

FX risk — — — Real estate risk -28.3 percent Up to −40 percent -25 percent Interest rate risk For USD risk-free curve: between -21 bps for

one year and -143 bps for 30 years+250 bps/-100 bps parallel shift Short-term rates -75 percent,

long-term rates -20 percent Liabilities Life underwriting Mortality/morbidity/longevity Pandemic: +1.5 additional deaths per 1,000 n/a Sensitivity analysis: decrease in

mortality

(continued)



496



Country United States Norway Germany

Year (FSSA publication) 2015 2015 2016 Lapse/surrender rates — n/a Sensitivity analysis: increase in

lapse rates Non-life underwriting Natural catastrophe (based on PML) Three events: 1) Florida hurricane (similar to

“Andrew,” 1992, but as 1/250 year event); 2) California earthquake (similar to “Northridge,” 1994, but as 1/250 year event); 3) series of EF5

tornados in the US Midwest

Two catastrophic events with claims of NOK 5 billion each

—

Other non-life underwriting shocks — — — Other risks Potential loss from guarantees in variable

annuities: 10 percent of the maximum guaranteed amount

— —

Risk aggregation/diversification effects Simple summation Simple summation and Solvency II correlation matrix

Correlation matrix

5. Regulatory Capital StandardsEffect on shareholder equity Buffer capital utilization (inverse of the

solvency capital requirement) with and without transitionals

Solvency capital requirement with and without transitionals

6. Presentation of Results Business segment/peer group breakdown Split into life/non-life/health/credit — Segmentation for large, medium,

small insurers Dispersion measures Interquartile distribution for effect on shareholder

equity— Interquartile distribution for effect

on solvency ratio Contribution of individual shocks Capital impact of individual shocks Capital impact of individual shocks Capital impact of individual shocks Other — — —

Source: Jobst, Sugimoto, and Broszeit 2014.Note: This table is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit. ABS = asset-backed security; bps = basis points; BU = bottom up; CHF = Swiss Franc; CMBS = commercial mortgage-backed secutiry; DKK = Danish Krona; EF5 = Enhanced Fujita scale rate of 5 (>200 mph); EUR = euro; FX = foreign exchange; FSAP = Financial Sec-tor Assessment Program; GAAP = Generally Accepted Accounting Principles; GBP = UK Pound Sterling; LGD = loss-given-default; max = maximum; min = minimum; P&C = property & casualty; PD = probability of default; PML = probable maximum loss; RBC = risk-based capital; RMBS = residential mortgage-backed security; SCR = solvency capital requirement; SE = South-East; TD = top down; QIS5 = Fifth Quantitative Impact Study; ZAR = South African rand.1 This table was originally presented in IMF Working Paper 14/133 (Jobst, Sugimoto, and Broszeit 2014).2 Based on gross premium income.3 Based on insurance liabilities.4 Only policies where lapses result in losses.



Andreas A



497


Stress Testing Matrix (STeM): Insurance Stress Testing Approaches in National Supervisory Frameworks (as of February 2014)1

Country Austria Bermuda Canada Czech Republic Denmark European Union2

1. Scope Approach BU BU BU BU BU BU Coverage Entire sector All large commercial (re)

insurers [Classes 4 and 3B]Entire sector (but small

insurers as well as term life and unit-linked insurers can

be exempted)

Large and middle-sized insurers

Entire sector Major European (re)insurance entities

(200 firms)

Relevance of the coverage

100 percent 100 percent n/a 90 percent of gross premium written

100 percent 50 percent of gross premium written

Reporting basis Solo Solo/consolidated n/a Solo n/a Consolidated (excluding banking activities)

Frequency Semi-annual for life and health insurers, annual

for non-life insurers

Annual Annual Annual Annual Annual

2. Valuation Basis Assets Statutory accounting Statutory accounting Statutory accounting Statutory accounting3 Statutory accounting Statutory accounting Liabilities Statutory accounting Statutory accounting Statutory accounting Statutory accounting3

(rough estimation of the change in the deficiency

provision in life insurance)

Statutory accounting Market-based approach of best estimate of liability projected into financial

statements

3. Scenario Design Source Provided by supervisor;

prescriptive shocksProvided by supervisor;

prescriptive shocksMostly principles-based with

some prescriptive shocksProvided by supervisor;



prescriptive shocks Macro-financial

linkage/transmis-sion channel(s)

Combination of single- (instantaneous) factor

shocks


shocks

n/a Combination of single- (instantaneous) factor

shocks


shocks


shocks Risk horizon Single period (stress is

assumed to occur at the end of a one-year

horizon)

Single period Multiple periods (five years [life], three years

[non-life])

Single period Single period Single period

Confidence level n/a 99.0 percent CTE n/a n/a n/a n/a

(continued)



498




4. Reg. Capital StandardsSolvency I Bermuda Solvency Standard Minimum Continuing Capital

and Surplus RequirementsSolvency I Solvency I Solvency II/Swiss Solvency

Test (SST)

5. General CommentContains no macro- fi-nancial specifications

and amounts to a sensitivity analysis

High comprehensiveness on technical (underwriting) risks; sensitivity analysis

exercises similar to Solvency II tests

Approach relies on dynamic financial analysis (DFA)

completed by firms, which use DFA techniques to model the uncertainty of insurance

operations [including scenarios and subsequent responses (second order

effects)]

Contains macro-financial linkages of insurance and

capital market shocks; additional features can be incorporated in one-year risk horizon, such as the

profit/loss produced during the year, the repeated occurrence of natural

disasters, and planned dividend payments

“Traffic-light” system with a yellow and a red

scenario; missing the thresholds of either

scenario is directly linked to heightened supervi-

sory scrutiny; yellow scenario suspended since

2008:Q3

Contains no macro- finan-cial specifications and

amounts to a sensitivity analysis; Switzerland

conducted stress tests together with EIOPA based on the same scenarios (but

the SST) for capital assessment

6. OutputPoststress effect on

solvency ratio(full impact/full impact net of hidden reserves/

full impact net of hidden reserves and

the equalization reserve)

Poststress effect on statutory assets and liabilities

Statutory ratio poststress either positive or above minimum depending on

scenario

Poststress effect on Solvency I ratio and the

ability to cover technical provisions with a sufficient volume of assets; economic

view to the interest rate sensitivity of assets and

liabilities

Solvency ratio poststress Reduction of own funds and comparison to MCR

(continued)


Andreas A



499




7. Risk Factors Macroscenario No specification of

general macroeco-nomic conditions

No specification of general macroeconomic

conditions

No specification of a baseline scenario; inflation

considered only for P&C

Fully fledged macroeco-nomic model linking

macro variables to capital market parameters; two

adverse scenarios (depression and loss of

confidence)

No specification of general macroeco-nomic conditions

Baseline and adverse scenarios with 0 percent

inflation change and a single inflationary

scenario

Assets Credit risk Yes

(-5 percent for “A-” to “BBB-”; -20 percent for

non-IG bonds)

Yes(rating class-specific increase

of credit spreads)

Yes Yes — Yes

Equity risk Yes(up to -35 percent)

Yes(-40 percent)

Yes(-35 percent, +15 percent,

-10 percent and level thereafter)

Yes Yes(yellow: -30 percent, red:

-12 percent)

Yes(up to -15 percent)

FX risk — Yes(up to -20 percent relative to

major currencies)

— Yes n/a —

Real estate risk Yes(-20 percent)

— — Yes Yes(up to -12 percent)

Yes(up to 11.6 percent for

residential, up to 25 percent for commercial)

Interest rate risk Yes(up to -10 percen)

— 3-month T-bill (-10 bps), long-term (-170 bps)

Yes Yes(yellow: +/- 100 bps, red:

+/- 70 bps)

Yes

Sovereign risk — Yes4 — Yes Government bond spread DKK-DEU

Yes

Other assets Hedge funds -40 percent

— — — — —

(continued)



500




Liabilities Life underwriting Mortality/

morbidity/longevity

— Yes(life loss is included in Lloyd’s

RDS for P&C)

Yes n/a — Yes(maximum exposure to mortality and longevity

shocks) Pandemic — Yes — — — — Lapse risk — — — — — — Reinsurance n/a — Yes — — — Other risk

factors— — Yes

(expense persistency, cash flow mismatch, and new

business [renewal])

— — —


Natural catastrophe

Yes Yes(Lloyd’s RDS scenarios for

P&C/own worst case scenarios)

Yes(implicit in internal models

underpinning DFA approach)

Yes(frequency and severity of

natural catastrophes)

— Yes(maximum exposure to two specific 1/200 year

scenarios [with reinsurance allowed to be included

discounted by 70 percent]; inflation shock to claims

reserves) Reinsurance n/a — Yes — — — Other risk

factorsYes

(higher frequency of claims in various lines

of business)

Yes(terrorism)

Yes(frequency, severity,

reinsurance, premium volume, misestimating

liabilities, pricing)

Yes(premium risk for casco/

motor third-party liability insurance, shock to net

written premiums following a macroeconomic model)

— —

(continued)


Andreas A



501




Other business Health insurance (being similar to life

insurance): increase in claims by 7.5 percent

Off-balance sheet items; rating downgrade (up to

two notches)

Regulatory and political risks Off-balance sheet items(look-through approach for

market and credit risk)

— —

Other risk factors Second-order

effects— — Yes

(managerial and regulatory reaction)

— — Yes(management actions)

Combination of financial/underwriting scenarios

Yes Yes Yes(implicit in internal models


Yes — Yes

Risk mitigation (reinsurance and

hedging)

n/a Yes(completion with and

without hedging assumption)



Yes(completion with existing

hedging assumptions)

n/a Yes(completion with and

without hedging assumption)


Various aggregations of stresses, generally no diversification effects

Limited(approximation of peak

exposure based on combined impact from

the three largest underwriting risks)



Aggregation by simple summation, no diversifica-

tion effects applied

n/a Limited(only inflation impact on

P&C scenarios)

(continued)



502



Country Germany Guernsey Japan Singapore Switzerland United Kingdom United States (NAIC)

1. Scope Approach1 BU/TD BU BU BU BU BU BU Coverage Most insurance firms

(but small insurers may be exempted)

6 life with liabilities >GBP 50 mln and 22 non-life

firms with gross premium earned >GBP 15 mln;

includes cell companies

All insurers, (re)insurers and

branches

All insurers All insurers Major life insurers All life and health (re)insurers

Relevance of the coverage

88 percent5 n/a 100 percent 100 percent 100 percent n/a 100 percent

Reporting basis Solo n/a n/a Solo/consolidated Solo/consolidated/granular

Solo/consolidated Solo

Frequency Annual (TD), quarterly (BU)6

Ad hoc n/a Annual Annual/semiannual Annual Annual

2. Valuation Basis Assets Statutory accounting Statutory accounting n/a Statutory accounting Market-consistent Statutory

accountingStatutory accounting

Liabilities Statutory accounting Statutory accounting n/a Statutory accounting Market-consistent Statutory accounting

Statutory accounting

3. Scenario Design Source Provided by

supervisor; prescrip-tive shocks

Provided by supervisor; prescriptive shocks

General guidelines but principle-based

approach; historical and hypothetical shocks/scenarios

General guidelines but principle-based

approach

Provided by supervisor; prescriptive shocks,

plus company-specific scenarios defined by insurance companies

Provided by supervisor; prescriptive

(standardized) shocks

Deterministic scenarios prescribed

by regulator; stochastic scenarios

generated by prescribed scenario

generator Macro-financial

linkage/transmission channel(s)


shocks


shocks


shocks


shocks

Combination of (instantaneous)

multiple factor shocks

Combination of (instantaneous) multiple factor

shocks

Combination of single- (instantaneous) factor shocks (interest rates/

market returns)

(continued)


Andreas A



503



Country Germany Guernsey Japan Singapore Switzerland United Kingdom United States (NAIC) Risk horizon Single period Single period Not specified Single period and

multiple period (3 years)

Single period Single period Over lifetime of liabilities

Confidence level n/a n/a n/a n/a Probability between 0.1 and 1 percent for

prescriptive scenarios

n/a 70 percent CTE for reserves and 90

percent CTE for capital

4. Reg. Capital StandardsGerman Solvency I Minimum Capital

Requirement (of licensed insurers)

Solvency Margin Ratio Singapore Risk- Based Capital

Swiss Solvency Test (SST)

Individual Capital Adequacy Standards

Risk-based Capital (RBC)

5. General CommentContains no macro- financial specifica-tions (other than

market risk shocks) and amounts to a sensitivity analysis

Simple model with selected single-factor

shocks

Publicly available information limited to

stress testing; approach is heavily

reliant on firm- generated stress

scenarios

Publicly available information limited to

stress testing for life insurers; approach is

heavily reliant on internal models and

firm-generated stress scenarios

High comprehensive-ness on technical

(underwriting) risks

Aimed to evaluate the resilience of

major life insurance groups to market

stresses of progressive severity

in order to understand the

nature of the market risks that the

groups are exposed to

Stress testing is done as test on the

adequacy of statutory formula reserves; at

start of projection, it is assumed that assets equal liabilities (e.g.,

surplus is zero); metric is the present value of surplus at end of the

projection period.

6. OutputAsset coverage ratio

over liabilitiesMargin of solvency has to

be higher than the minimum margin of

solvency

n/a Risk-based capital ratio poststress

Aggregation of several scenarios to the

capital requirement (target capital)

Poststress impacts Reserves adequacy (negative surplus

requires additional reserves)

(continued)



504




7. Risk Factors Macroscenario No specification of

general macroeco-nomic conditions

Two scenarios, baseline and adverse for the

downward movement of interest rates and rate of

inflation affecting non-life claims volumes

Included, but no consistent specifica-

tion of general macroeconomic

conditions across firms

No specification of general macroeconomic

conditions


No specification, but scenarios combining

substantial equity, property, credit, and

yield shifts


Assets Credit risk Yes — Yes

(firm-specific scenarios according to

supervisory guidelines [applied to all other

sub-categories below])

Yes(firm-specific

scenarios over a one-year risk horizon according to supervisory guidelines

[applied to all other sub-categories below])

Yes(credit spreads only)

Yes(spread for

corporate bonds)

Yes(rating class-specific

increase of credit spreads)

Equity risk Yes7 Yes Yes Yes Yes(stochastic)

FX risk — Yes Yes — Yes Real estate risk Yes

(-10 percent)8

Yes Yes Yes Yes

Interest rate risk Yes9 Yes Yes Yes Yes(deterministic and

stochastic) Sovereign risk — Yes

(overlaps with interest rate scenarios)

— — Yes(credit spreads only)

— —

Other assets — — — Yes

(continued)


Andreas A



505




Liabilities Life underwriting Mortality/

morbidity/Longevity

— Yes Yes(firm-specific scenarios

according to supervisory

guidelines [applied to all other


Yes(firm-specific scenarios over a

three-year risk horizon according to

supervisory guidelines [applied to all other


Yes — Yes(sensitivity tested)

Pandemic — Yes Yes — — Lapse risk — — Yes — Yes

(sensitivity tested) Reinsurance — — Yes — — Other risk

factorsYes

(bond and equity test: decline in the price

of equities and 5 percent decline in

the price of fixed-income

securities; equity and property test: decline in the price of equities and 10 percent decline in

the market value of properties)

Yes(expense persistency)

Yes — Expenses and other policyholder

behavior


— n/a

Natural catastrophe

— Yes(with and without reinsurance default)

Yes(firm-specific scenarios

according to supervisory guidelines)

Yes(firm-specific scenarios

over a three-year risk horizon according to

supervisory guidelines)

— — n/a

Reinsurance — Yes Yes — n/a Other risk

factors— Yes

(claims inflation)Yes — n/a

Other business — — — — Yes — n/a

(continued)



506




Other risk factors Second-order

effects— — — Yes

(management actions)Yes

(management actions)Yes

(managerial and regulatory action)

n/a

Combination of financial/underwriting scenarios

— — Yes(firm-specific

scenarios)

Yes(firm-specific scenarios)

Yes — n/a

Risk mitigation (reinsurance and

hedging)

Yes(only financial risks)

Yes(only reinsurance)

Yes(firm-specific

scenarios)

Yes(firm-specific scenarios)

Yes(for insurance risk

computed gross and/or net of reinsurance)

— Yes


Yes(only financial risks)10

— — — Risk factors are shocked simultaneously

— Yes

Sources: BaFin; Bank of England (PRA); Bermuda Monetary Authority; Czech National Bank; IMF; National Association of Insurance Commissioners; and Swiss Financial Market Supervisory Authority. Note: This table is available on the IMF eLibrary at https://www.elibrary.imf.org/page/stress-test2-toolkit. casco = acronym built from the first letters of the words casualty and collision—motor casco insurance covers risks that could lead to partial damage or complete loss of a car; CTE = conditional tail expectation; EIOPA = European Insurance and Occupational Pensions Authority; DKK-DEU = Danish Krona-Deutsche Mark exchange rate; IG = investment grade-related; MCR = minimum capital requirement; NAIC = National Association of Insurance Commissioners; P&C = property & casualty; PRA = Prudential Regulatory Au-thority; RBC = risk-based capital; RDS = Realistic Disaster Scenarios; Reg = regulatory.1This table was originally presented in IMF Working Paper 14/133 (Jobst, Sugimoto, and Broszeit 2014).2Also Switzerland contributes to the EIOPA stress test based on the same specification and scenarios (but applies the Swiss Solvency Test (SST) for capital assessment).3In addition, the economic view to the interest rate risk of assets and liabilities is applied. 4The sovereign risk shock is considered from a creditor perspective by examining the potential magnitude of both valuation changes and impairment charges of mark-to-market (MTM) and hold-to-maturity (HTM) assets. Haircuts are calculated from expected valuation changes of liquid government (benchmark) bonds, assuming an increase of sovereign distress but not a general shift in the yield curve. The ap-proach follows Jobst and Oura 2019.5Number of submissions vs. number of supervised solo undertakings in 2013.6Annually, with obligation to report to supervisor; internally, the undertakings are obliged to perform the test quarterly.7In order to avoid explicit procyclical behavior, the equity shock is based on a rules-based system depending on the year-end value of the Euro Stoxx 50. In 2012, the shock was 18 percent for the equity only scenario.8But only in combination with equity price shock.9Market value reduction of 10 percent of fixed income instruments according to an assumed interest rate rise. For the low interest rate environment, BaFin uses a different approach termed “scenario calculation.”10Only in combined scenarios, with credit risk calculated (shocked) in every scenario.


https://www.elibrary.imf.org/page/stress-test2-toolkit.%20casco%20=%20acronym%20built%20from%20the%20first%20letters%20of%20the%20words%20casualty%20and%20collision%E2%80%94motor%20casco%20insurance

Appendix 17.2.National and IMF Stress Testing

for Non- Life (Re)insurance: A Case Study of Bermuda

Bermuda is host to one of the largest insurance market in the world, with globally active, commercial underwriters focused on property and casualty risks. The supervisory stress testing framework is an important component of the solvency regime for these firms. All large firms are required to perform an annual stress test as specified by the Bermuda Monetary Authority (BMA) and submit the results with their Bermuda Capital and Solvency Return.

The stress testing exercise aims at assessing the capital adequacy of the legal entities and groups by evaluating the impact of risk drivers conditional on plausible scenarios defined by firm- and system- wide changes. It is designed to provide a comprehen-sive understanding of the general loss- absorbing capacity of insurers in relation to the economic impact of shocks to asset prices and interest rates as well as projected losses arising from specific underwriting risks on the insurer’s or group’s statutory balance sheet (that is, statutory admitted assets, admitted liabilities, and capital and surplus). The BMA requires stress testing to be conducted at the firm level, either as part of firms’ internal models or using prescriptive shocks to risk factors in accordance with uniform guidelines and assumptions. It is based on either internal model- derived or predefined scenarios affecting both the single entity and group- wide annual solvency return.1

The annual stress testing exercise examines the impact of a rapid deterioration of both investment and underwriting perfor-mance over a wide range of scenarios to assess the expected impact and effects of adverse events on statutory assets and liabili-ties. The financial impacts reported for underwriting scenarios are those that would be observed immediately upon the occurrence of the event as determined by the firm’s internal or vendor model(s) (both with and without the effect of reinsurance and/or other loss mitigation instruments):

• Economic scenarios (financial risk): Asset risks include several capital market- related single- factor shocks triggered by an adverse global macroeconomic scenario: (1) a severe decline in equity prices (40 percent), without allowance for diversi-fication across the markets (that is, assuming that all markets are correlated and only long (asset) positions are affected); (2) a widening of credit spreads;2 (3) a negative exchange rate shock to all and net open foreign currency positions (that is, assuming a depreciation of the US dollar versus major reserve currencies);3 and (4) valuation haircuts on fixed-income holdings (and long derivatives positions) of sovereign debt and financial bonds (debt securities and loans), including a general upward shift in the yield curve of 50 basis points.4

• Underwriting scenarios (insurance risk): Multiple underwriting risks affecting aggregates in force at the beginning of the reporting period are examined based on: (1) prescribed property and casualty events in different scenario groupings (US windstorm, US earthquake, non- US windstorm, non- US earthquake, aerospace/aviation, and marine) as specified in Lloyd’s Handbook on Realistic Disaster Scenarios; (2) nonpeak perils, which do not currently exist in vendor models (US oil spill, US tornadoes, Australian flooding, and Australian wildfires); (3) additional insurance risks (pandemic,

1 The stress test of underwriting performance allows firms to also consider own worst- case scenarios as a substitute for the prescriptive scenarios. In this regard, the BMA encourages firms to apply internal stress testing approaches that utilize in- house expertise in risk management and the data/models to facilitate the management of important risk drivers according to their own risk appetite and risk profile.

2 In the exercise for the 2013 reporting year (BMA 2013a and 2013b), credit spreads widen across different rating classes (from 163 basis points for “AAA”-rated securities, to 3,188 basis points for securities rated “BB” or lower). The adverse scenario (“ through- the- cycle”) is calibrated to historical price changes of rating- specific baskets of benchmark credit default swap (CDS) with three- year maturity at a statistical confidence of 99th percentile.

3 The magnitude of negative shock for each currency (euro, Japanese yen, pound sterling, Swiss franc, and Australian dollar, with the US dollar as the refer-ence currency) is determined based on four times the difference between the maximum implied annualized volatility (1) between January 1, 2008 and the end of 2011, and (2) the long- term average since January 1, 2005.

4 The sovereign- risk shock comprises valuation changes and impairment charges in economic terms and are applied to all net exposures (that is, gross expo-sures net of cash [short] positions [without derivatives hedges], including both on- and off- balance sheet assets and claims irrespective of accounting treatment [ mark- to- market, available-sale, and hold- to- maturity]). These haircuts are calculated from expected valuation changes of liquid government (benchmark) bonds based on changes in sovereign risk implied by the 99th percentile of the historical density of one-, three-, five-, seven-, and 10-year forward contracts on CDS with maturity terms between one and 10 years following Jobst and Oura 2019.



terrorism); (4) other underwriting scenarios (if the prescribed perils either do not apply or partially apply to the insurer/group resulting in de minimis loss projections); (5) projections from the worst- case annual aggregate catastrophe loss scenario, which combines economic and underwriting loss scenarios generating the largest losses and a series of loss simulations relating to extreme tail events;5 and (6) a qualitative assessment of a credit rating downgrade by two notches (or falling below a “ A-” rating, whichever is more severe) on income and liquidity positions.6 All lines of business and ex-posures are included in the final estimates of the loss impact net of protection, such as reinsurance, retrocessional agree-ments, or insurance- linked securities.

The reporting firms submit a description of all key assumptions and calculations as well as the positions on aggregate statu-tory assets and liabilities that would be observed immediately upon the occurrence of various shocks (both with and with-out the effect of reinsurance and/or other loss mitigation instruments). The results also comprise both the occurrence return period (for example, a one- in- 50-year event) and the relative return period (that is, using the underlying loss distribution of the aggregate net probable maximum loss to calculate the corresponding return period of each event).

In 2007, the IMF completed a BU system- wide solvency stress test of 10 large commercial property and casualty as well as long- term insurance companies as part of the Offshore Financial Center Assessment Program (IMF 2008b). The companies employed a combination of their respective internal and vendor models to calculate the capital impact of a variety of underwrit-ing scenarios (three natural catastrophe events, two pandemic events, and worst- case scenarios of aggregate net probable maxi-mum loss as specified by each insurer). The impact of these scenarios was assessed against statutory reporting requirements at the time (that is, change in capital and surplus, minimum regulatory- premium ratio, and minimum regulatory- loss- reserve ratio). The results of the stress test exercise suggested that catastrophic events would have had a significantly negative impact on aggregate capital, with the most severe impact resulting from the worst- case scenarios, only two of which included economic events (in addition to natural disasters). Scenarios combining catastrophic events and an economic recession had the greatest impact on the solvency positions of firms on average. No firm failed to meet applicable regulatory capital requirements under any of the scenarios; however, the exercise only covered a subset of risk factors considered necessary to attain a comprehensive assessment of the risk profile of firms under stress.7

5 More specifically, each firm is to submit the results of the aggregate impact of (1) a combination of a financial market scenario (assuming only a severe decline in equity prices and a widening of credit spreads) and an aggregation of the three largest net underwriting losses, and (2) either a series of loss simulations or results of other analyses performed related to extreme tail events or a firm- specific, worst- case, annual aggregate loss scenario at a level considered extreme but plausible by the firm.

6 The disclosure should cover and provide an indication of the relative impact and severity of collateral requirements, loss payment triggers on in- force policy contracts, claw- backs, and/or other adverse financial and liquidity implications of the downgrade.

7 The incomplete coverage of financial market shocks and use of accounting data rather than economic valuation were identified as shortcomings of the supervisory stress testing framework, which were remedied in the first stress testing guidance as part of the newly established risk- based solvency regime the BMA introduced in 2010. In particular, the sensitivity of some firms to the combined impact of several extreme financial and underwriting events motivated the introduction of a worst- case annual aggregate catastrophe loss.



Stability Review 2013. Frankfurt am Main, Germany: Deutsche Bundesbank. https://www.bundesbank.de/en/publications/reports /financial-stability-reviews/financial-stability-review-2013 -654260.

Dionne, Georges, and Kili C. Wang. 2013. “Does Insurance Fraud in Automobile Theft Insurance Fluctuate with the Business Cy-cle?” Journal of Risk and Uncertainty 47 (1): 67–92. https://www .jstor.org/stable/43550176?seq=1#page_scan_tab_contents.

European Central Bank (ECB). 2013. Financial Stability Review 2013. Frankfurt am Main, Germany: European Central Bank. https://www.ecb.europa.eu/pub/pdf/fsr/f inancialstability re view201311en.pdf?4af555f0d4cf971eb0efb60474bf6fa0.

European Insurance and Occupational Pension Authority (EIOPA). 2011a. “EIOPA Announced Today the Results of Its Second European Insurance Stress Test.” July 4 Press Release issued by the European Insurance and Occupational Pension Authority, Frankfurt am Main, Germany.

———. 2011b. “Specification for the 2011 EU- Wide Stress Test in the Insurance Sector.” EIOPA- FS- 11/012, European Insurance and Occupational Pension Authority, Frankfurt am Main, Ger-many. https://eiopa.europa.eu/ financial- stability- crisis- prevention / financial- stability/ insurance- stress-test/insurance-stress-test-2011.

———. 2013a. “Opinion on Supervisory Response to a Prolonged Low Interest Rate Environment.” EIOPA- BoS 12/110, European Insurance and Occupational Pension Authority, Frankfurt am Main, Germany.

———. 2013b. “Financial Stability Report, Second Half- Year Re-port, Autumn 2013.” European Insurance and Occupational Pension Authority, Frankfurt am Main, Germany. https://eiopa .europa.eu/ financial- stability- crisis- prevention/ financial- stability / financial- stability- reports/ second- half-year-financial-stability -report-2013.

———. 2014. “EIOPA Insurance Stress Test 2014.” European In-surance and Occupational Pension Authority, Frankfurt am Main, Germany. https://eiopa.europa.eu/ financial- stability - crisis- prevention/ f inancialstability/ insurance- stress-test /insurance-stress-test-2014.

———. 2016. “EIOPA Insurance Stress Test Report.” European In-surance and Occupational Pension Authority, Frankfurt am Main, Germany. https://eiopa.europa.eu/Pages/ Financial- stability-and -crisis-prevention/Stress-test-2016.aspx.

———. 2018. “Technical Documentation of the Methodology to Derive EIOPA’s Risk- free Interest Rate Term Structures.” Eu-ropean Insurance and Occupational Pension Authority, Frank-furt am Main, Germany.

European Systemic Risk Board (ESRB). 2015. “Report on Systemic Risks in the EU Insurance Sector.” European Central Bank, Frankfurt am Main, Germany. https://www.esrb.europa.eu /pub/reports/html/index.en.html?69f998814350279530108928e0188b97.

———. 2016. “Macroprudential Policy beyond Banking: An ESRB Strategy Paper.” European Systemic Risk Board, Frank-furt am Main, Germany. https://www.esrb.europa.eu/news/pr /date/2016/html/pr160719.en.html.

Financial Services Agency of Japan (JFSA). 2013. “Comprehensive Guidelines for the Supervision of Insurance Companies.” Fi-nancial Services Agency of Japan, Tokyo. https://www.fsa.go .jp/en/news/2014/20141209-1.html.

Financial Services Authority (FSA). 2008. “Stress and Scenario Testing.” Consultation Paper CP 08/24, Financial Services Au-thority, London, UK.

REFERENCESAbdymomunov, Azamat, Sharon Blei, and Bakhodir Ergashev.

2011. “ Worst- Case Scenarios as a Stress Testing Tool for Risk Models.” Working Paper, The Federal Reserve Bank of Rich-mond, Charlotte Branch, June 21.

Antolin, Pablo, Sebastian Schich, and Juan Yermo. 2011. “The Eco-nomic Impact of Protracted Low Interest Rates on Pension Funds and Insurance Companies.” OECD Journal: Financial Market Trends 2001 (1): 1–20. https://www. oecd- ilibrary.org/ finance - and- investment/ the- economic- impact- of- protracted- low - interest- rates- on- pension- funds-and-insurance-companies _fmt-2011-5kg55qw0m56l.

Artzner, Philippe, Freddy Delbaen, Jean- Marc Eber, and David Heath. 2001. “Coherent Measures of Risk.” Mathematical Finance 9 (3): 203–28. https://onlinelibrary.wiley.com/doi /pdf/10.1111/1467-9965.00068.

Bermuda Monetary Authority (BMA). 2013a. 2013 Capital and Solvency Return: Stress/Scenario Analysis— Class 4, Class 3B and Insurance Groups. Hamilton: Bermuda Monetary Authority.

———. 2013b. 2013 Capital and Solvency Return: Stress/Scenario Analysis— Class 3A. Hamilton: Bermuda Monetary Authority.

———. 2013c. Bermuda Insurance- Linked Securities (ILS)Market Report: Vol. 1, No. 1. Hamilton: Bermuda Monetary Authority.

Borio, Claudio, Matthias Drehmann, and Kostas Tsatsaronis. 2012. “ Stress- Testing Macro Stress Testing: Does It Live Up to Expectations?” BIS Working Paper 369, Bank for Interna-tional Settlements, Basel, Switzerland. https://www.bis.org /publ/work369.htm.

Breuer, Thomas, Martin Jandačka, Javier Mencía, and Martin Summer. 2012. “A Systematic Approach to Multi- Period Stress Testing of Portfolio Credit Risk.” Journal of Banking and Fi-nance 36 (2): 332–40. https://www.sciencedirect.com/science /article/pii/S0378426611002202.

Bundesanstalt für Finanzdienstleistungsaufsicht (BaFin). 2004. “Conducting of Stress Test.” Circular 1/2004 (VA), Bunde-sanstalt für Finanzdienstleistungsaufsicht, Bonn, Germany. http://www.bafin.de/DE/Aufsicht/VersichererPensionsfonds /Stresstest/stresstest_node.html.

Committee of European Insurance and Occupational Pensions Su-pervisors (CEIOPS). 2009. “Building a European Stress Test for the European Insurance Sector.” CEOPS- FSC- 31/09, Com-mittee of European Insurance and Occupational Pensions Su-pervisors, Frankfurt am Main, Germany.

———. 2010. “Results of CEIOPS EU- Wide Stress Test for the Insurance Sector.” March 16, Press Release issued by the Com-mittee of European Insurance and Occupational Pensions.

Committee on the Global Financial System (CGFS). 2000. “Stress Testing by Large Financial Institutions: Current Practice and Aggregation Issues.” CGFS Publication 14, Bank for Interna-tional Settlements, Basel, Switzerland. https://www.bis.org /publ/cgfs14.htm.

———. 2005. “Stress Testing at Major Financial Institutions: Sur-vey Results and Practice.” CGFS Publication 24, Bank for In-ternational Settlements, Basel, Switzerland. https://www.bis .org/publ/cgfs24.htm.

———. 2012. “Operationalizing the Selection and Application of Macroprudential Instruments.” CGFS Paper 48, Bank for In-ternational Settlements, Basel, Switzerland. https://www.bis .org/publ/cgfs48.htm.

Deutsche Bundesbank. 2013. “Insurance Companies: Bridging Low Interest Rates and Higher Capital Requirements.” In Financial


https://www.bundesbank.de/en/publications/reports/financial-stability-reviews/financial-stability-review-2013-654260



https://www.jstor.org/stable/43550176?seq=1#page_scan_tab_contents

https://www.jstor.org/stable/43550176?seq=1#page_scan_tab_contents

https://www.ecb.europa.eu/pub/pdf/fsr/financialstability





https://eiopa.europa.eu/Pages/

https://www.esrb.europa.eu/pub/reports/html/index.en.html?69f99881435027953010892


https://www.esrb.europa.eu/news/pr/date/2016/html/pr160719.en.html

https://www.esrb.europa.eu/news/pr/date/2016/html/pr160719.en.html

https://www.fsa.go.jp/en/news/2014/20141209-1.html

https://www.fsa.go.jp/en/news/2014/20141209-1.html

https://onlinelibrary.wiley.com/doi/pdf/10.1111/1467-9965.00068

https://onlinelibrary.wiley.com/doi/pdf/10.1111/1467-9965.00068



https://www.sciencedirect.com/science/article/pii/S0378426611002202

https://www.sciencedirect.com/science/article/pii/S0378426611002202

http://www.bafin.de/DE/Aufsicht/VersichererPensionsfonds/Stresstest/stresstest_node.html

http://www.bafin.de/DE/Aufsicht/VersichererPensionsfonds/Stresstest/stresstest_node.html

https://www.bis.org/publ/cgfs14.htm.






https://www. oecd- ilibrary.org/ finance- and- investment/ the- economic- impact- of- protracted- low- interest- rates- on- pension- funds-and-insurance-companies_fmt-2011-5kg55qw0m56l.







https://eiopa.europa.eu/Pages/


https://www.ecb.europa.eu/pub/pdf/fsr/financialstability


———. 2012a. “Global Systemically Important Insurers: Pro-posed Policy Measures.” Public Consultation Document, Bank for International Settlements, Basel, Switzerland.

———. 2012b. “Reinsurance and Financial Stability.” Bank for International Settlements, Basel, Switzerland. https://www .iaisweb.org/file/34046/reinsurance-and-financial-stability.

———. 2012c. “Global Insurance Market Report (GIMAR).” Bank for International Settlements, Basel, Switzerland. https://w w w.ia isweb.org/page/news/ g loba l- insurance- market - report- gimar/f ile/41508/global-insurance-market-report -2012-edition/.

———. 2012d. “Global Systemically Important Insurers: Proposed Assessment Methodology.” Public Consultation Document, Bank for International Settlements, Basel, Switzerland.

———. 2013a. “Macroprudential Policy and Surveillance in Insur-ance.” Macroprudential Surveillance and Policy Subcommit-tee, Bank for International Settlements, Basel, Switzerland. https://www.iaisweb.org/file/34258/mps-report-18-july-2013.

———. 2013b. “Global Systemically Important Insurers: Final Initial Assessment Methodology.” Bank for International Set-tlements, Basel, Switzerland. https://www.iaisweb.org/file/34257 / final- initial-assessment-methodology-18-july-2013.

———. 2013c. “Global Systemically Important Insurers: Final Policy Measures.” July 18 Press Release issued by Bank for International Settlements, Basel, Switzerland. https://www .iaisweb.org/page/ supervisory- material/ financial- stability- and - macroprudential- policy- and- surveillance/file/34256/ final- g- siis -policy-measures-18-july-2013.

——— 2013d. “Basic Capital Requirements (BCR) for Global Systemically Important Insurers ( G- SIIs).” Bank for Interna-tional Settlements, Basel, Switzerland. https://www.iaisweb .org/file/34540/ iais- basic-capital-requirements-for-g-siis.

———. 2015. “Global Systemically Important Insurers: Proposed Updated Assessment Methodology.” Public Consultation Doc-ument, Bank for International Settlements, Basel, Switzerland. ht tps://w w w.ia i sweb.org /f i le /58005/ g- s i i-a s se s sment -methodology-public-consultation-document.

———. 2018a. “Draft Overall ComFrame.” Public Consultation, Bank for International Settlements, Basel, Switzerland. https://w w w.ia isweb.org/page/ super v isor y- mater ia l / common - framework/file/76143/ draft-overall-comframe-for-public -consultation.

———. 2018b. “ICS Version 2.0.” Public Consultation, Bank for International Settlements, Basel, Switzerland. https://www .iaisweb.org/page/ supervisory- material/ insurance- capital- standard /file/76133/ ics-version-20-public-consultation-document.

International Monetary Fund (IMF). 2003. “Japan: Financial Sys-tem Stability Assessment and Supplementary Information.” IMF Country Report 03/287, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/30/ Japan -Financial-System-Stability-Assessment-16865.

———. 2004a. “The Kingdom of the Netherlands: Financial Sys-tem Stability Assessment.” IMF Country Report 04/312, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/ The- K ingdom- of- the- Netherlands - Netherlands- Financia l-System-Stabi l it y-A ssessment -including-17756.

———. 2004b. “France: Financial System Stability Assessment.” IMF Country Report 04/344, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ France - Financia l- Sy s t em- S t abi l i t y- A s s e s sment- i nc lud ing - Reports-on-the-Observance-of-17818.

———. 2009. “Stress and Scenario Testing: Feedback on Consul-tation Paper CP 08/24 and Final Rules.” Policy Statement 9/20, London, UK. https://docplayer.net/21670236- Policy - statement- 09-20- f inancial- services- authority- stress- and - scenario- testing- feedback-on-cp08-24-and-final-rules.html.

French, Doug, Richard de Haan, Robb Luck, and Justin Mosbo. 2011. “The Impact of Prolonged Low Interest Rates on the In-surance Industry.” Ernst & Young, New York.

Geneva Association. 2010a. “Systemic Risk in Insurance—An Analysis of Insurance and Financial Stability: Special Report of the Geneva Association Systemic Risk Working Group.” The International Association for the Study of Insurance Econom-ics, Geneva, Switzerland. https://www.genevaassociation.org / research- topics/ financial- stability- and- regulation/ systemic -risk-insurance-analysis-insurance-and.

———. 2010b, “Key Financial Stability Issues in Insurance—An Account of the Geneva Association Ongoing Dialogue on Sys-temic Risk with Regulators and Policy- Makers.” The Interna-tional Association for the Study of Insurance Economics, Geneva, Switzerland. https://www.genevaassociation.org/ research- topics / f inancialstability- and- regulation/key-financial-stability -issues-insurance.

———. 2012. “Cross Industry Analysis: 28 G- SIBs vs. 28 Insurers Comparison of Systemic Risk Indicators.” The International Association for the Study of Insurance Economics, Geneva, Switzerland. https://www.genevaassociation.org/ research- topics / financial- stability- and- regulation/ cross- industry-analysis28 -g-sibs-vs-28-insurers.

Guernsey Financial Services Commission. 2012. “Stress Testing of the Guernsey Insurance Sector.” Guernsey Financial Services Commission, St Peter Port, Guernsey. https://www.gfsc.gg /news/article/stress-testing-october-2012.

Hesse, Heiko, Salman Ferhan, and Christian Schmieder. 2014. “How to Capture Macro- Financial Spillover Effects in Stress Tests?” IMF Working Paper 14/103, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/ How- to- Capture- Macro- Financial - Spillover-Effects-in-Stress-Tests-41644.

International Actuarial Association (IAA). 2013. “Stress Testing and Scenario Analysis.” International Actuarial Association, Ottawa, Canada.

International Association of Insurance Supervisors (IAIS). 2003. “Stress Testing by Insurers.” Guidance Paper 8, October, Bank for International Settlements, Basel, Switzerland. https://www .iaisweb.org/page/ supervisory- material/ archive- supervisorarchive - supervisory- material- superseded- by- icps- standards- guidance - adopted- in- 2011/ guidance- papers/file/34131/8- stress- testing - by-insurers-guidance-paper-october-2003.

———. 2010. “Macroprudential Surveillance and (Re)Insurance.” Global Reinsurance Market Report, Mid- year Edition, Bank for International Settlements, Basel, Switzerland. https://www .iaisweb.org/page/ supervisory- material/ financial- stability- and - macroprudential- policy- and- surveillance/ global- insurance - market- report- gimar//f ile/34782/ iais- global- reinsurance - market-report-2010-mid-year-edition.

———. 2011a. “Insurance and Financial Stability.” Bank for In-ternational Settlements, Basel, Switzerland. https://www.iaisweb .org/f ile/34706/15- november- 2011- iais- issues- paper-on -insurance-and-financial-stability.

———. 2011b. “Insurance Core Principles, Standards, Guidance and Assessment Methodology.” Bank for International Settle-ments, Basel, Switzerland.


https://www.iaisweb.org/file/34046/reinsurance-and-financial-stability.

https://www.iaisweb.org/file/34046/reinsurance-and-financial-stability.

https://www.iaisweb.org/page/news/


https://www.iaisweb.org/file/34258/mps-report-18-july-2013.

https://www.iaisweb.org/file/34257/

















https://docplayer.net/21670236-Policy-

https://www.genevaassociation.org/


https://www.gfsc.gg/news/article/stress-testing-october-2012

https://www.gfsc.gg/news/article/stress-testing-october-2012







https://www.iaisweb.org/file/34706/15-november-





























https://www.genevaassociation.org/financial- stability- and- regulation/ cross- industry-analysis28-g-sibs-vs-28-insurers.




———. 2007e. “Switzerland: Financial Sector Assessment Program— Technical Note on Insurance Sector Stress Testing.” IMF Country Report 07/201, Washington, DC. https://www.imf .org/en/Publications/CR/Issues/2016/12/31/ Switzerland - Financial- Sector- Assessment- Program- Technica l-Note -Insurance-Sector-Stress-21051.

———. 2007f. “Mexico: Financial Sector Assessment Program Update— Technical Note on Risk Management Practices and Stress Tests of Commercial Banks, the Insurance Sector, and the Derivatives Exchange.” IMF Country Report 07/165, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/ Mexico- Financial- Sector- Assessment - Program- Update- Technical-Note-Risk-Management-Practices -20964.

———. 2008a. “South Africa: Financial System Stability Assess-ment.” IMF Country Report 08/349, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ South- Africa - Financia l- System- St abi l it y- A s s e s sment- Inc lud i ng - Report-on-the-Observance-of-22437.

———. 2008b. “Bermuda: Assessment of the Supervision and Regulation of the Financial Sector.” IMF Country Report 08/336, Washington, DC. https://www.imf.org/en/Publica tions /CR /Is sues/2016/12/31/ Bermuda- A sse s sment- of- the - Supervision- and- Regulation-of-the-Financial-Sector-22421.

———. 2008c. “Austria: Financial Sector Assessment Program Update— Technical Note on Factual Update and Analysis of the IAIS Insurance Core Principles.” IMF Country Report 08/207, Washington, DC. https://www.imf.org/en/Publica tions /C R / I s s u e s /2 016/12 /31/ A u s t r i a - F i n a nc i a l - S e c to r - Assessment- Program- Update- Technical-Note-Factual-Update -and-22119.

———. 2009a. “Isle of Man: Financial System Stability Assess-ment.” IMF Country Report 09/275, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ Isle- of - Man- Financia l- Sector- A ssessment- Program- Update -Financial-System-Stability-Assessment-23266.

———. 2009b. “Isle of Man: Financial Sector Assessment Pro-gram Update— Technical Note on Stress Testing Banking and Insurance.” IMF Country Report 09/279, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / Isle- of- Man- Financial- Sector- Assessment- Program- Update - Technical-Note-Stress-Testing-Banking-23270.

———. 2010a. “United States: Financial System Stability Assess-ment.” IMF Country Report 10/247, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ United - States- Publication- of- Financial- Sector-Assessment-Program -Documentation-Financial-24105.

———. 2010b. “United States: Financial Sector Assessment Pro-gram Update— Technical Note on Stress Testing.” IMF Coun-try Report 10/244, Washington, DC. https://www.imf.org/en /Publications/CR/Issues/2016/12/31/ United- States- Publication - of- Financial- Sector-Assessment-Program-Documentation -Technical-24101.

———. 2011a. “Guernsey: Financial System Stability Assessment— Update.” IMF Country Report 11/1, Washington, DC. https://w w w. i m f .o r g /e n / P u b l i c a t i on s /C R / I s sue s /2016/12 /31/ Guernsey- Financial-System-Stability-Assessment-Update -24542.

———. 2011b. “Guernsey: Financial Sector Assessment Program Update— Technical Note on Stress Testing: Banking and In-surance.” IMF Country Report 11/4, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ Guernsey

———. 2004c. “Singapore: Financial System Stability Assess-ment.” IMF Country Report 04/104, Washington, DC. https://w w w.imf.org /en/Publ icat ions/CR /Issues/2016/12/30 / Singapore-Financial-System-Stability-Assessment-17332.

———. 2005. “France: Financial Sector Assessment Program— Technical Notes on Stress Testing Methodology and Results.” IMF Country Report 05/185, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ France - Financial- Sector- Assessment- Program- Detailed- Assessments -of-Observance-of-Standards-18300.

———. 2006a. “Belgium: Financial System Stability Assessment.” IMF Country Report 06/75, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ Belgium - Financia l- System- St abi l it y- A s s e s sment- i nc lud i ng - Reports-on-the-Observance-of-18956.

———. 2006b. “Spain: Financial System Stability Assessment.” IMF Country Report 06/212, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ Spain- Financial - System- Stabi l it y- A ssessment- inc lud ing- Repor t s- on -the-Observance-of-Standards-19336.

———. 2006c. “Spain: Financial Sector Assessment Program— Technical Note on Stress Testing Methodology and Results.” IMF Country Report 06/216, Washington, D.C. https://www .imf.org/en/Publicat ions/CR /Issues/2016/12/31/ Spain - Financial- Sector- Assessment- Program- Technical- Note-Stress -Testing-Methodology-and-19340.

———. 2006d. “Denmark: Financial System Stability Assess-ment.” IMF Country Report 06/343, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ Denmark - Financial- System- Stability- Assessment- including- Reports-on -Observance-of-Standards-19933.

———. 2006e. “Mexico: Financial System Stability Assessment.” IMF Country Report 06/350, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ Mexico - Financial- System- Stability- Assessment- Update- including -Summary-Assessments-on-the-19978.

———. 2006f. “Portugal: Financial System Stability Assessment.” IMF Country Report 06/378, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ Portugal - Financial- System- Stability- Assessment- including- Reports -on-the-Observance-of-20034.

———. 2007a. “Portugal: Financial Sector Assessment Program— Technical Note on Stress Testing.” IMF Country Report 07/34, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/ Portugal- Financial- Sector- Assessment - Program-Technical-Note-Stress-Testing-20347.

———. 2007b. “Denmark: Financial Sector Assessment Program— Technical Note on Stress Testing.” IMF Country Report 07/125, Washington, DC. https://www.imf.org/en/Publications/CR /Issues/2016/12/31/ Denmark- Financial- Sector- Assessment - Program-Technical-Note-Stress-Testing-20601.

———. 2007c. “Denmark: Financial Sector Assessment Program— Detailed Assessment of Observance of the Insurance Core Principles.” IMF Country Report 07/119, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / Denmark- Financial- Sector- Assessment- Program-Technical -Note-Stress-Testing-20601.

———. 2007d. “Switzerland: Financial System Stability Assess-ment Update.” IMF Country Report 07/187, Washing-ton, DC. https://www.imf.org/en/Publications/CR/Issues /2016/12/31/ Switzerland- Financial-System-Stability-Assessment -Update-21022.


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/South- Africa- Financial- System- Stabi l it y- A s se s sment- Inc luding- Report-on-the-Observance-of-22437

https://www.imf.org/en/Publica/CR/Issues/2016/12/31/ Bermuda- A ssessment- of- the- Supervision- and- Regulation-of-the-Financial-Sector-22421

https://www.imf.org/en/Publica/CR/ I s sue s /2016/12/31/ A u s t r i a - F inanci a l - S e c tor- Assessment- Program- Update- Technical-Note-Factual-Update





















































https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Switzerland-Financial-Sector-Assessment-Program-Technical-Note-Insurance-Sector-Stress-21051




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Mexico-Financial-Sector-Assessment-Program-Update-Technical-Note-Risk-Management-Practices-20964




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Isle-of-Man-Financial-Sector-Assessment-Program-Update-Financial-System-Stability-Assessment-23266




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/United-States-Publication-of-Financial-Sector-Assessment-Program-Documentation-Financial-24105




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Mexico-Financial-System-Stability-Assessment-Update-including-Summary-Assessments-on-the-19978




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Portugal-Financial-System-Stability-Assessment-including-Reports-on-the-Observance-of-20034




https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Portugal-Financial-Sector-Assessment-Program-Technical-Note-Stress-Testing-20347



https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Denmark-Financial-Sector-Assessment-Program-Technical-Note-Stress-Testing-20601




/2016/12/31/ Canada- Financial- Sector- Assessment- Program -Stress-Testing-Technical-Note-41405.

———. 2014c. “Denmark: Financial System Stability Assess-ment.” IMF Country Report 14/336, Washington, D.C. https://w w w.imf.org /en/Publ icat ions/CR /Issues/2016/12/31 / Denmark-Financial-System-Stability-Assessment-42504.

———. 2014d. “Denmark: Financial Sector Assessment Program— Technical Note on Stress Testing the Banking, Insurance, and Pension Sectors.” IMF Country Report 14/348, Wash-ington, DC. https://www.imf.org/en/Publications/CR/Issues /2016/12/31/ Denmark- Stress- Testing- the- Banking- Insurance - and-Pension-Sectors-Technical-Note-42536.

———. 2014e. “South Africa: Financial System Stability Assess-ment.” IMF Country Report 14/340, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ South - Africa-Financial-System-Stability-Assessment-42508.

———. 2015a. “South Africa: Financial Sector Assessment Program— Technical Note on Stress Testing the Financial Sys-tem.” IMF Country Report 15/54, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ South - A fr ica- Financia l- Sector- A ssessment- Program- Stress -Testing-the-Financial-System-42756.

———. 2015b. “United States: Financial System Stability Assess-ment.” IMF Country Report 15/170, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ United - States- Financial- Sector- Assessment- Program-Financial -System-Stability-Assessment-43055.

———. 2015c. “United States: Financial Sector Assessment Program— Technical Note on Stress Testing.” IMF Country Report 15/173, Washington, DC. https://www.imf.org/en /Publications/CR/Issues/2016/12/31/ United- States- Financial - Sector- Assessment- Program-Stress-Test ing-Technica l -Notes-43058.

———. 2015d. “Norway: Financial System Stability Assessment.” IMF Country Report 15/252, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ Norway - F i n a nc i a l - S e c tor- A s s e s sment- Prog r a m-Fi n a nc i a l -System-Stability-Assessment-43263.

———. 2015e. “Norway: Financial Sector Assessment Program— Technical Note on Insurance Sector Stress Tests.” IMF Coun-try Report 15/255, Washington, DC. https://www.imf.org/en /Publications/CR/Issues/2016/12/31/ Norway- Financial- Sector - Assessment- Program- Technical- Note-Insurance-Sector-Stress -Tests-43268.

———. 2016a. “Germany: Financial System Stability Assess-ment.” IMF Country Report 16/189, Washington, DC. https://w w w.imf.org /en/Publ icat ions/CR /Issues/2016/12/31 / Germany- Financial- Sector- Assessment- Program-Financial -System-Stability-Assessment-44013.

———. 2016b. “Germany: Financial Sector Assessment Program— Technical Note on Stress Testing the Banking and Insurance Sectors.” IMF Country Report 16/191, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / Germany- Financial- Sector- Assessment- Program- Stress - Testing- the-Banking-and-Insurance-Sectors-44015.

Ionescu, Liviu, and Juan Yermo. 2014. “Stress Testing and Sce-nario Analysis of Pension Plans.” IOPS Working Paper on Ef-fective Pensions Supervision No.19, Organisation for Economic Cooperation and Development (OECD)—International Or-ganisation of Pension Supervisors, Paris.

Jobst, Andreas A. 2012. “Systemic Risk in the Insurance Sector—A Review of General Issues and Some Findings on Large

- Financial- Sector- Assessment- Program- Update- Technical - Note-on-Stress-Testing-Banking-24545.

———. 2011c. “Macroprudential Policy: An Organizing Frame-work.” Monetary and Capital Markets, International Monetary Fund, Washington, DC.

———. 2011d. “Toward Operationalizing Macroprudential Poli-cies: When to Act?” Chapter 3. Washington, DC, September. https://www.imf.org/~/media/Websites/IMF/imported-flag ship -issues/external/pubs/ft/GFSR/2011/02/pdf/_ch3pdf.ashx.

———. 2011e. “Luxembourg: Financial System Stability Assessment— Update.” IMF Country Report 11/148, Washington, DC. https://w w w.imf.org/en/Publicat ions/CR /Issues/2016 /12/31/ Luxembourg- Financial-System-Stability-Assessment -Update-24995.

———. 2012a. “Israel: Financial System Stability Assessment.” IMF Country Report 12/69, Washington, DC. https://www .imf.org/en/Publicat ions/CR /Issues/2016/12/31/ Israel -Financial-System-Stability-Assessment-25815.

———. 2012b. “ Macro- Financial Stress Testing Principles and Practices.” Monetary and Capital Markets Department, Inter-national Monetary Fund, Washington, DC. https://www.imf .org/en/Publicat ions/ Policy- Papers/Issues/2016/12/31 / Macrofinancial- Stress-Testing-Principles-and-Practices-PP4702.

———. 2012c. “Israel: Financial Sector Assessment Program— Technical Note on Stress Testing of the Banking, Insurance and Pension Sectors.” IMF Country Report 12/88, Washing-ton, DC. https://www.imf.org/en/Publications/CR/Issues/2016 /12/31/Israel-Technical-Note-on-Stress-Test-of-the- Banking -Insurance-and-Pension-Sectors-25850.

———. 2012d. “Japan: Financial System Stability Assessment— Update.” IMF Country Report 12/210, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ Japan - Financial-Sector-Stability-Assessment-Update-26137.

———. 2012e. “France: Financial System Stability Assessment.” IMF Country Report 12/341, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ France -Financial-System-Stability-Assessment-40187.

———. 2013a. “Belgium: Financial System Stability Assessment.” IMF Country Report 13/124, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ Belgium -Financial-System-Stability-Assessment-40547.

———. 2013b. “Belgium: Financial Sector Assessment Program— Technical Note on Stress Testing the Banking and Insurance Sectors.” IMF Country Report 13/137, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31 / Belgium- Technical- Note- on- Stress- Testing- the-Banking-and -Insurance-Sectors-40573.

———. 2013c. “Key Aspects of Macroprudential Policy.” IMF Policy Paper, Washington, DC. https://www.imf.org/en /Publications/ Policy- Papers/Issues/2016/12/31/ Key-Aspects -of-Macroprudential-Policy-PP4803.

———. 2013d. “Singapore: Financial System Stability Assess-ment.” IMF Country Report 13/325, Washington, DC. https://www.imf.org/en/Publications/CR/Issues/2016/12/31/ Singapore -Financial-System-Stability-Assessment-41051.

———. 2014a. “Canada: Financial Sector Stability Assessment.” IMF Country Report 14/29, Washington, DC. https://www .imf.org/en/Publications/CR/Issues/2016/12/31/ Canada -Financial-Sector-Stability-Assessment-41299.

———. 2014b. “Canada: Financial Sector Assessment Program— Technical Note on Stress Testing.” IMF Country Report 14/69, Washington, DC. https://www.imf.org/en/Publications/CR/Issues






















https://www.imf.org/en/Publications/CR/Issues/2016/12/31// Germany- Financial- Sector- Assessment- Program- Stress- Testing- the-Banking-and-Insurance-Sectors-44015

https://www.imf.org/~/media/Websites/IMF/imported-flag

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Luxembourg- Financial-System-Stability-Assessment


https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Israel

https://www.imf.org/en/Publications/.org/en/Publications/ Policy- Papers/Issues/2016/12/31/Macrofinancial- Stress-Testing-Principles-and-Practices-PP4702

https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Israel-Technical-Note-on-Stress-Test-of-the-
















https://www.imf.org/en/Publications/CR/Issues






https://www.imf.org/en/Publications/CR/Issues/2016/12/31/Israel























https://www.imf.org/~/media/Websites/IMF/imported-flag


from 16 January 2012],” Notice No. MAS 312, Monetary Au-thority of Singapore, Singapore. February 7. http://www.mas .gov.sg/ regulations- and- financial- stability/ regulations- guidance - and- licensing/insurance/notices/ insurance- companies/2011 / mas- 312- stress- testing- on- financial- condition- of-life-direct -insurer_7-feb-2011.aspx.

———. 2013. “Enterprise Risk Management for Insurers.” Notice No. MAS 126, Monetary Authority of Singapore, Singapore, April 2. http://www.mas.gov.sg/ Regulations- and- Financial - Stability/ Regulations- Guidance- and- Licensing/Insurance /Notices/ Insurance-Companies/2015/MAS-126_7-Dec-2015.aspx.

National Association of Insurance Commissioners (NAIC). 2013. “Own Risk and Solvency Assessment (ORSA).” Presented at 2013 NAIC Financial Summit, San Diego, May 29. https://www.naic.org/Releases/2013_docs/naic_financial_summit _2013_leadership_initiatives.htm.

Office of the Superintendent of Financial Institutions (OSFI). 2009. “Stress Testing.” Guideline, Government of Canada, Of-fice of the Superintendent of Financial Institutions, Ottawa. http://www. osfi- bsif.gc.ca/Eng/ fi- if/ rg-ro/gdn-ort/gl-ld/Pages /e18.aspx.


Prudential Regulatory Authority (PRA), 2013. “The Prudential Regulation Authority’s Approach to Insurance Supervision.” Bank of England, London. https://www.bankofengland.co .uk /-/media /boe/f i les/prudentia l-regulat ion/approach /insurance-approach-2016.

Swiss Federal Office of Private Insurance. 2004. “White Paper of the Swiss Solvency Test.” Swiss Federal Office of Private Insur-ance, Bern, Switzerland.

Swiss Re. 2012. “Facing the Interest Rate Challenge.” Sigma No. 4/2012, Swiss Re, Zurich, Switzerland.

Vouldis, Angelos, Patrizia Baudino, Christoffer Kok, and Matthias Sydow. 2013. “Box 4: Insurance Sector Solvency Analysis Framework: A Stock- Taking of Available Tools.” In A Macro Stress Testing Framework for Assessing Systemic Risks in the Bank-ing Sector, Occasional Paper Series No. 152, edited by Jérôme Henry and Christoffer Kok. Frankfurt: European Central Bank. https://www.ecb.europa.eu/pub/research/authors/profiles /jerome-henry.en.html.

Wang, Lin, and Ali M. Kutan. 2013. “The Impact of Natural Di-sasters on Stock Markets: Evidence from Japan and the United States.” Comparative Economic Studies 55: 672–86. https://link .springer.com/article/10.1057%2Fces.2013.16.

Insurers in Bermuda.” Working Paper, Bermuda Monetary Au-thority, Hamilton, Bermuda. https://www.researchgate.net /publication/256015445_Systemic_Risk_in_the_Insurance _Sector_-_A_Review_of_General_Issues_and_Some_Findings _on_Large_Insurers_in_Bermuda.

———. 2013a. “Best Practices in Insurance Stress Testing.” Insur-ance Industry Forum: ORSA, Section 3: Group Risk Capital and Prospective Solvency Assessment, Institute for Interna-tional Research (IIR), December 11, Boston, Massachusetts.

———. 2013b. “Multivariate Dependence of Implied Volatilities from Equity Options as Measure of Systemic Risk.” Interna-tional Review of Financial Analysis 28: 112–29. https://papers .ssrn.com/sol3/papers.cfm?abstract_id=2194213##.

———. 2014. “Systemic Risk in the Insurance Sector: A Review of Current Assessment Approaches.” The Geneva Papers on Risk and Insurance 39: 440–70. https://link.springer.com/article /10.1057/gpp.2013.7.

Jobst, Andreas A., Li Lian Ong, and Christian Schmieder. 2013. “A Framework for Macroprudential Bank Solvency Stress Test-ing: Application to S- 25 and Other G20 Country FSAPs.” IMF Working Paper 13/68, International Monetary Fund, Wash-ington, DC. https://www.imf.org/en/Publications/WP/Issues /2016/12/31/ A- Framework- for- Macroprudential- Bank- Solvency - Stress- Testing- Application- to- S-25-and-Other-G-40390.

Jobst, Andreas A., and Hiroko Oura. 2019. “Sovereign Risk in Bank Solvency Stress Testing.” IMF Working Paper, International Mon-etary Fund, Washington, DC.

Jobst, Andreas A., Nobuyasu Sugimoto, and Timo Broszeit. 2014. “Macroprudential Solvency Stress Testing of the Insurance Sec-tor.” IMF Working Paper 14/133, International Monetary Fund, Washington, DC. https://www.imf.org/en/Publications /WP/Issues/2016/12/31/ Macroprudential- Solvency- Stress - Testing-of-the-Insurance-Sector-41776.

Kablau, Anke, and Michael Wedow. 2012. “Gauging the Impact of a Low- Interest Rate Environment on German Life Insurers.” Applied Economics Quarterly 58 (4): 279–98. Also published as Discussion Paper 02/2011, Deutsche Bundesbank, Frankfurt am Main, Germany. https://elibrary.duncker-humblot.com /journals/id/22/vol/58/iss/1630/art/6923/.

Komárková, Zlatuše, and Marcela Gronychová. 2012. “Models for Stress Testing in the Insurance Sector.” CNB Research and Policy Notes 2, Czech National Bank, Prague. https://www .cnb.cz/en/research/research_publications/irpn/2012/rpn_02 _2012.html.

Monetary Authority of Singapore (MAS). 2011. “Stress Testing on Financial Condition of Life Direct Insurer [Cancelled with effect


http://www.mas.gov.sg/.gov.sg/ regulations- and- financial- stability/ regulations- guidance- and- licensing/insurance/notices/ insurance- companies/2011

http://www.mas.gov.sg/Regulations- and- Financial- Stability/ Regulations- Guidance- and- Licensing/Insurance/Notices/ Insurance-Companies/2015/MAS-126_7-Dec-2015.aspx

https://www.naic.org/Releases/2013_docs/naic_financial_summit_2013_leadership_initiatives.htm



http://www.osfi- bsif.gc.ca/Eng/ fi- if/ rg-ro/gdn-ort/gl-ld/Pages

http://www.mdpi.com/2227-7072/2/1/15

https://www.bankofengland.co.uk/-/media/boe/files/prudential-regulation/approach/insurance-approach-2016



https://www.ecb.europa.eu/pub/research/authors/profiles/jerome-henry.en.html

https://www.ecb.europa.eu/pub/research/authors/profiles/jerome-henry.en.html

https://link.springer.com/article/10.1057%2Fces.2013.16

https://link.springer.com/article/10.1057%2Fces.2013.16

https://www.researchgate.net/publication/256015445_Systemic_Risk_in_the_Insurance_Sector_-_A_Review_of_General_Issues_and_Some_Findings_on_Large_Insurers_in_Bermuda.






https://link.springer.com/article/10.1057/gpp.2013.7

https://link.springer.com/article/10.1057/gpp.2013.7



https://elibrary.duncker-humblot.com/journals/id/22/vol/58/iss/1630/art/6923/

https://elibrary.duncker-humblot.com/journals/id/22/vol/58/iss/1630/art/6923/

https://www.cnb.cz/en/research/research_publications/irpn/2012/rpn_02_2012.html








http://www.osfi- bsif.gc.ca/Eng/ fi- if/ rg-ro/gdn-ort/gl-ld/Pages





Page numbers with f, t, or b indicate figures, tables, or boxes respectively.

Accounting-based modelsfor FSAP solvency stress tests, 389, 391ffor sovereign risk, 225

Acharya, Viral, 157Action Plan for Strengthening Surveillance, by

IMF, 88Adrian, Tobias, 58Advanced economies (AEs)

credit growth in, 130, 133f, 138credit loss rates in, 126, 126n25, 127fimpairment income in, 130fleverage ratio in, 140LGDs in, 129, 129tNPL in, 124preimpairment income in, 129–30, 130freal GDP growth in, 134, 135n44retained earnings in, 138ROC in, 122–23solvency tests for, 121, 121n6sovereign risk in, 183spillover effects in, 286stress tests of, 273–74

Advanced internal-ratings-based (AIRB), RWAs, 243–45, 245f, 248f, 250, 253, 257t

Afonso, Gara, 157AfS. See Available-for-saleAgent-based models

for solvency risk, 4for stress tests, 59

Aggregation problem, for stress tests, 57–58Aikman, David, 5, 435AIRB. See Advanced internal-ratings-basedAlbania, 102, 109, 109n20Albertazzi, U., 308Altavilla, Carlo, 225Anand, Kartik, 435Antalya summit, of G20, 83n4Antifragility, 272n12, 273f, 274AQR. See Asset quality reviewArbitrage

for banking books, 259liquidity and, 416tmarket-consistent valuation and, 188sovereign risk and, 225for trading books, 259

Asset correlationsGDP growth and, 152fIRB and, 138–40, 139b, 139f

Asset encumbrance, of FSAPs liquidity stress tests, 420

Asset-liability matching, in insurance sector, 459Asset quality review (AQR)

in crisis stress tests, 323, 324, 338, 348–49, 362f, 365

in SCAP, 339, 350

Attractor loan default rate, GFM and, 69Australia

FSAP solvency stress tests in, 375GIIPS and, 288, 289fshocks in, 314

Austriainsurance FSAPs in, 486tring-fencing in, 102n6

Available-for-sale (AfS), 13, 14n4sovereign risk and, 187–88, 196transparency with, 196n19valuation gap of, 196n18

Avgouleas, Emilios, 103n9Aymanns, Christoph, 156n7

Babihuga, Rita, 157Bailouts

feedback loops for, 93by Greece, 329f, 356tby Ireland, 328, 329fsovereign risk and, 185, 225spillover effects and, 49f

Balance of Payments Manual, of IMF, 81Balance Sheet Analysis in Fund Surveillance, by

IMF, 92Balance sheets

aggregation problem and, 57CCA for, 203–4FSAP solvency stress tests and, 384, 389,

391fin insurance sector, 457f, 467n35, 480solvency tests and, 13–14, 13fstress test models, 19b, 20t, 57unconsolidated, 4See also Off-balance-sheet

Banco de España (BdE), 324, 347Bank credit

asset changes and, 50BIS and, 308destabilization of, 204GFM and, 67sensitivity of, 135in UK, 73

BankFocus, 292Bank for International Settlements (BIS), 12

bank credit and, 308cross-border banking and, 90, 308, 309GFM and, 75Handbook on Securities Statistics by, 88in IAG, 82n1IBS of, 82, 86, 90, 96liquidity risk and, 414tRTF of, 12, 414t

Bank groups, mapping of, 113, 113t–16tBanking books

arbitrage for, 259FSAP liquidity stress tests and, 422sovereign risk and, 188

Banking crisescredit growth in, 130, 133fcredit loss rates in, 125b, 126f, 126t, 129fLGD in, 128–29, 128f, 129tpreimpairment income in, 129–30, 130fsolvency tests in, 123–30, 124t, 125b,

125f–28f, 126t, 129t, 130f–33f, 132bBank of England, 251

Risk Assessment Model for Systemic Institutions of, 291b

Bankscope, 122, 126, 142cross-border banking and, 309data from, 147f, 147t, 148t

Barnhill, Theodore, 25b, 291b, 435Basel Committee on Banking Supervision

(BCBS), 3, 21n16, 377G-SIBs and, 87liquidity risk and, 414tPrinciples for Sound Stress Testing Practices

and Supervision of, 39RCAP of, 237–38, 238fon SIB, 380Sixth Quantitative Impact Study of, 387sovereign risk and, 188–89

Basel I, 14n7OECD and, 249n10RWAs and, 249, 249n10, 250, 259–61,

260t–61t, 261fsolvency stress tests and, 377

Basel I/IIcapital ratios in, 238, 238n3RWAs and, 242

Basel II, 14n5covered bonds and, 254b, 254tFSAP solvency stress tests and, 385microprudential/supervisory stress tests

and, 15RWAs and, 249, 259–61, 260t–61t, 261f,

344–45, 385solvency stress tests and, 377solvency tests and, 121stress tests of, 56

Basel II/III, 254ring-fencing and, 102, 103n8RWAs and, 243–44

Basel III, 14n4, 39Common Equity Tier I of, 47covered bonds and, 254bCT1 of, 159–61, 238, 239dividends and, 384FSAP solvency stress tests and, 385LCR of, 416, 419, 422, 431–32, 433tNSFR of, 416, 419, 422, 432RWAs and, 50, 253, 255, 387solvency stress tests and, 377solvency tests and, 121sovereign risk and, 184, 185b, 200transition schedule for, 386t

Index


Index516

Baudino, Patrizia, 3n2Bayoumi, Tamim, 274, 286n4BCBS. See Basel Committee on Banking

SupervisionBdE. See Banco de EspañaBeau, Emily, 158n11Bédard-Pagé, Guillaume, 435Behavioral cash flows, FSAPs liquidity stress

tests and, 421–22Belgium

CDS in, 197insurance FSAPs in, 469, 485tinsurance sector in, 477, 478bliquidity ratio of, 420shocks in, 313stress tests in, 478b

Benchmarksfor capital ratio, 30bfor cross-border banking, 105FSIs as, 85for housing loans, 281n26for liquidity, 416, 423MtM as, 29for PCAR, 328for real GDP growth, 139bfor sovereign risk, 187–89for spillover effects, 287for stress tests, 120for tail risks, 21

Bermuda, insurance FSAPs in, 515–16Bernanke, Ben, 55, 344Bese Goksu, Evrim, 3Best practices, for stress tests, 1–3, 21–36,

37tBhargava, Alok, 165BIS. See Bank for International SettlementsBlack, Fischer, 203n40Black swan, 21, 35–36, 36n35, 130

crisis stress test and, 340fragility and, 269, 269n1, 271

Borio, Claudio, 340Boston Consulting Group, 255n14Bottero, M., 308Bottom-up stress tests (BU), 17n11, 17t, 338,

340crisis-management stress tests and, 46FSAP liquidity stress tests as, 418FSAP solvency stress tests as, 379, 388, 392,

394t–98t, 454for insurance sector, 462in Spain, 347–48

Brazilcapital ratios in, 109n20GIIPS and, 289, 290fsovereign risk in, 183

Brownlees, Christian T., 157Brunnermeier, Markus K., 26, 58Bruno, Brunnella, 251n11BU. See Bottom-up stress testsBurden sharing agreements, 102

Caceres, Carlos, 421n10Calibration

of country-specific interest rate shocks, 214n62

of FSAP liquidity stress tests, 420of GFM, 75of macroprudential solvency stress tests,

189–96, 190t–94t

of shocks, 189–96, 190t–94tof sovereign risk, 185, 189–96, 190t–94tof stress tests, 31–34, 32b–33b

Call optionCCA and, 203n40Moody’s reports and, 159n16with strike price, 15n8

Canadacapital ratios in, 109n20FSAP solvency stress tests in, 471GIIPS and, 288, 289finsurance FSAPs in, 469, 486tOSFI in, 475bRWAs in, 248shocks in, 312, 314

Capital adequacy ratio (CAR), 187–88market valuation and, 189

Capital buffers, 20, 59, 394t–97tcountercyclical fiscal policies and, 146, 255crisis stress tests for, 321–23for cross-border banking, 307–9, 313, 314,

317, 318–19interest rates and, 28microprudential/supervisory stress tests

for, 15RWAs and, 238–39

Capital Exercise, of EU, 348Capital ratios, 109n20

in Basel I/II, 238, 238n3benchmark for, 30bconcentration risk and, 28GFM and, 67procyclicality and, 184, 225regulatory bank, 69RWAs and, 237, 238–41, 255solvency test for, 14sovereign risk and, 225stress tests for, 278

Capital Requirement Directive (CRD), 254b, 344

Capital requirements, 15, 27b, 56, 59, 383of EBA, 238of EU, 106funding cost and, 168GFM and, 70OTC derivatives and, 88for trading books, 250

Capital utilization rate, 66CAR. See Capital adequacy ratioCash flow

expected and potential, 421n13FSAP liquidity stress tests and, 421–22ICF, 416, 419, 437–38, 442fof insurance sector, 459

Cash-flow based liquidity stress tests, 447Catastrophe risk, 474CBI. See Central Bank of IrelandCCA. See Contingent claims analysisCCAR. See Comprehensive Capital Analysis

and ReviewCCPs. See Central counterpartiesCDIS. See Coordinated Direct Investment

SurveyCDMs. See Concentration and distribution

measuresCDS. See Credit default swapsCEBS. See Committee of European Banking

SupervisorsCentral Bank of Ireland (CBI), 324

Central counterparties (CCPs)FSAPs for, 5–6stress tests for, 24

Cerutti, Eugenio, 4, 308CET1. See Core Equity Tier 1Cetina, Jill, 158n10Cetorelli, Nicola, 102, 308China

GIIPS and, 289, 290fsovereign risk in, 183

Christiano, Lawrence, 64Čihák, Martin, 18, 23b, 28n23, 123, 124t,

290–91, 308, 422Claessens, Stijn, 308Closed-form expression, in stress tests,

272n14Committee of European Banking Supervisors

(CEBS), 324–25, 328, 339, 344Committee on Payment and Settlement

Systems (CPSS), 22n17Commodity price cycle, 187Common Equity Tier I, 47Common interest rate shock, 195, 195n13Communication

DGI and, 91–92in stress tests, 34–35

Comprehensive Capital Analysis and Review (CCAR), 272n13, 350, 373

Concavity (linearity) effects, of tail risks, 270, 271f, 271n8, 278n23

Concentration and distribution measures (CDMs), 86, 86n9

Concentration riskcapital ratios and, 28in insurance sector, 472RWAs and, 255

Conditional tail expectation (CTE), 211, 212, 213

Consolidated balance sheets, ring-fence liquidity, 4

Consolidated stress tests, ring-fencing and, 101–13

Contagion risk, 28, 195n14, 308in insurance sector, 473–74stress tests for, 270–71

Contemporaneous net government asset ratio, 69

Contingent claims analysis (CCA), 203–4, 203n40

Contractual cash flows, 421–22Convexity (nonlinearity) effects

fragility and, 270n4with shocks, 206n49of tail risks, 270, 271–72, 274, 274n17

Coordinated Direct Investment Survey (CDIS), of IMF, 90, 96

Coordinated Portfolio Investment Survey (CPIS), of IMF, 75, 82, 90, 96

Core Equity Tier 1 (CET1), 159n14European Bank for Reconstruction and

Development and, 177, 179fsovereign risk and, 196

Core Tier 1 ratio (CT1)of Basel III, 159–61, 238, 239CDS and, 166, 168crisis stress tests and, 345, 346bhurdle rates and, 346bfor RWAs, 241, 242fshort-term debt and, 168


Index 517

Correlations. See Asset correlations; Dynamic conditional correlation

Counterbalancing capacity, FSAPs liquidity stress tests and, 421

Countercyclical fiscal policiescapital buffers and, 146, 255sovereign risk and, 183n1

Counterparty risk, 19n15FSAP liquidity stress tests and, 422sovereign risk and, 186

Country-specific interest rate shocks, 195calibration of, 214n62

Covered bonds, 251, 254b, 254tCPIS. See Coordinated Portfolio Investment

SurveyCPSS. See Committee on Payment and

Settlement SystemsCRAs. See Credit rating agenciesCRD. See Capital Requirement DirectiveCredit constraints

DGI and, 89GFM and, 65

Credit default swaps (CDS), 84, 88contracts for, 209b–10bcredit risk and, 195n16credit risk premium and, 208, 211–14, 215f,

216fcrisis stress tests and, 324, 325, 326t, 328CT1 and, 166, 168EDF and, 162–64fair value model for, 197n27GEV for, 196–97, 202liquidity and, 200n38LLPs and, 160–61for market-consistent valuation, 186bmaturity of, 197n33OTC, 186n1pricing formula for, 209b–10bRWAs and, 177solvency risk and, 157, 158–59sovereign risk and, 185, 202

Credit growthin banking crises, 130, 133fFSAP solvency stress tests and, 384GDP growth and, 138, 274, 274n18real GDP growth and, 135satellite models for, 274stress tests for, 281

Credit lossesin downturn conditions, 120GDP growth and, 136–37, 274, 274n18in Iceland, 134n43ROC and, 140–42satellite models for, 274stress tests for, 281trading losses and, 132b

Credit loss rates, 125b, 126f, 126t, 129fGDP growth and, 136–37GFM and, 67real GDP growth and, 135satellite models for solvency tests and,

135–37Credit rating agencies (CRAs), 255Credit ratings, 159Credit risk

CDS and, 195n16crisis stress tests for, 343tof cross-border banking, 308in downturn condition, 14

in economic cycle, 456n12HtM and, 219in insurance sector, 457, 472, 476–77NPLs and, 195n13of RWAs, 248–49, 249f, 249t, 259, 387shocks, 4sovereign risk and, 189, 195stress test models for, 50

Credit risk premiumin adverse scenarios, 211–14, 215f, 216fCDS and, 211–14, 215f, 216fGFM and, 70haircuts for, 214, 214n63sovereign risk and, 185, 208

Credit spreadsCCA and, 204solvency risk and, 5

Crises. See Banking crises; European sovereign debt crisis; Global financial crisis

Crisis-management stress tests, 15–17, 45–47, 460b

BU and, 46in GFC, 3See also Supervisory Capital Assessment

ProgramCrisis stress tests

AQR in, 323, 324, 338, 348–49, 362f, 365

for capital buffers, 321–23capital standards for, 344–45case studies of, 355, 356t–63tcredibility of, 321–65data for, 324–25, 324t–26tdesign of, 328–53design scorecard for, 334t–36tdisclosure of technical details of, 348, 349tDIV in, 323effectiveness scorecard for, 332tin EU, 323–24, 324t, 325t, 329f, 338,

340–41, 347, 351t–52t, 355, 356t–63tfollow-up tests for, 350–53forward-looking techniques for, 344governance of, 338hurdle rates and, 346bin Ireland, 323–24, 324t, 325t, 355,

356t–63tjurisdiction financial markets statistics for,

337tmacro-financial parameters scorecard for,

342trestructuring costs with, 353risk factors scorecard for, 343tscenario design for, 340–44, 341b, 341fscope of, 339–40for SIB, 339for solvency, 321–23, 322fin Spain, 323–24, 324t, 325t, 340, 341–44,

347–48, 347b, 355, 356t–63tstandardization of assumptions for, 344success of, 325–28timing of, 333, 338transparency of, 345–48in Turkey, 323in United States, 323–24, 324t, 325t,

351t–52t, 355, 356t–63tCroatia, 109, 109n20Crockett, Andrew, 56Cross-border banking

benchmarks for, 105capital buffers for, 307–9, 313, 314, 317,

318–19credit risk of, 308data on, 308–9, 310t–11tDGI and, 89–91, 96FSAP solvency stress tests and, 384funding risk of, 308GDP growth and, 5, 307–8, 316GDP surprises and, 316–17, 316n3, 316tGFC and, 307methodologies for, 309–12, 312frobustness of, 317–18, 318f, 319tshocks and, 312–15, 313f–16fspillover effects, 5ofvulnerabilities of, 307–19

CT1. See Core Tier 1 ratioCTE. See Conditional tail expectationCurrency redenomination, 183Currency risk, 90

in insurance sector, 477premium shocks, 66

Current account balance ratioGFM and, 69in GIIPS, 287

Czech Republic, 102, 109, 109n20insurance FSAPs in, 486t

Dagher, Jihad, 143Data Gap Initiative (DGI), of G20, 81–97

CDIS and, 96communication and, 91–92CPIS and, 96cross-border banking and, 96cross-country banking and, 89–91FSIs and, 85–86GFC and, 82–93GFF and, 84GFSM and, 96GFSR and, 82–83, 83fG-SIBs and, 83, 87G-SIFIs and, 87property markets and, 91, 97Public Sector Debt Database and, 97recommendations, 95–97regulatory reform agenda of, 83–85for risk management, 85–86, 85fsectoral analysis for, 88–89for shadow banks, 86–87stress tests and, 92–93surveillance agenda for, 84, 88–89

Data integrity and verification (DIV), 323DCC. See Dynamic conditional correlationDebt sustainability analysis, 185Default rate

attractor loan, 69GDP growth and, 137, 137fGFM and, 70sovereign risk and, 183

Deficits-to-GDP ratio, 287De Haas, Ralph, 102, 308Demographic risk, 474Denmark

FSAP solvency stress tests in, 471insurance FSAPs in, 485t, 486tinsurance sector of, 477

Derivatives, 84, 88, 160DFA. See Dodd-Frank ActDGI. See Data Gap Initiative


Index518

Dimensional dependence of maximum loss, 58n1

Discount rate, 203insurance sector and, 469

Distinguin, Isabelle, 157DIV. See Data integrity and verificationDodd-Frank Act (DFA), 17n12, 24n20, 246,

251, 321, 350Domestic demand, GFM and, 65, 68Douady, Raphael, 270n4, 272n15Downturn conditions, 19b, 39, 485b

credit losses in, 120credit risk in, 14for LGDs, 195, 195n12PDs in, 242spillover effects in, 285

Drehmann, Mathias, 340DSGE. See Dynamic stochastic-general

equilibriumDüllmann, Klaus, 138Duration risk premium, 66–73Durbin-Wu-Hausman test, 165–66, 168Dutch Central Bank, 25bDynamic conditional correlation (DCC)

GARCH and, 288–89, 297GFC and, 288n13for GIIPS, 288–89

Dynamic stochastic-general equilibrium (DSGE), 63, 189

Early Warning Exercise, of IMF, 373EBA. See European Banking Association;

European Banking AuthorityEC. See European CommissionECB. See European Central BankEconomic cycle

credit risk in, 456n12liquidity risk in, 456n12RWAs and, 247sovereign risk and, 183n1

EDF. See Expected default frequencye-GDDS. See Enhanced-GDDSEichenbaum, Martin, 64EIOPA. See European Insurance and

Occupational Pensions AuthorityEisenbeis, Robert A., 103n9Emerging market and developing economies

(EMDEs), 18sovereign risk in, 183, 185, 187, 195, 200,

225Emerging market economies (EMEs)

credit growth in, 130, 133f, 138credit loss rates in, 126, 126n25, 127fGDP growth in, 124impairment income in, 130fJ.P. Morgan Emerging Markets Bond Index

for, 287, 288leverage ratio in, 140LGDs in, 129NPL in, 124preimpairment income in, 129–30, 130freal GDP growth in, 134, 135n44retained earnings in, 138ROC in, 122–23solvency tests for, 121, 121n6spillover effects in, 286

Engle, Robert, 157Enhanced-GDDS (e-GDDS), 91Enrico, Luca, 84

Enterprise risk management, 456Equity risk, in insurance sector, 472,

487t–515tEquity risk premium

GFM and, 70shocks, 66, 72t, 73

Espinosa-Vega, Marco, 5, 308EU. See European UnionEuropean Bank for Reconstruction and

Development, 102CET1 and, 177, 179f

European Banking Association (EBA), 324European Banking Authority (EBA), 47f, 102,

111, 111t, 238–39European Central Bank (ECB), 47f, 344

crisis stress tests and, 328DGI and, 96Handbook on Securities Statistics by, 88in IAG, 82n1insurance sector and, 462n30LTRO of, 294, 328, 329f, 353quantitative easing by, 189SSM of, 350

European Commission (EC), 47fEuropean Financial Stability Facility, 347European Insurance and Occupational

Pensions Authority (EIOPA), 462, 475b, 480, 480n63

European peripheral countries (GIIPS)DCC for, 288–89GARCH for, 288–89, 289f, 292–93spillover effects of, 287–89, 289f, 290f

European sovereign debt crisis, 4, 103, 164, 328

haircuts in, 189lessons from, 18–21sovereign risk in, 183–84spillover effects in, 285, 287, 289

European Stability Mechanism, 347European Systemic Risk Board, 456n8European Union (EU)

Capital Exercise of, 348capital requirements of, 106CEBS of, 324–25, 328, 339, 344crisis stress tests in, 323–24, 324t, 325t,

329f, 338, 340–41, 347, 351t–52t, 355, 356t–63t

FSAP solvency stress tests for, 383insurance FSAPs in, 469, 486tmacroprudential solvency stress tests for,

196–200, 198t–99tring-fencing in, 105–10, 106f, 107t–9tSolvency I in, 467Solvency II in, 467, 480n62sovereign debt in, 286

EurostatDGI and, 96in IAG, 82n1

EU system-wide stress test, 45, 45n46, 45n48, 47f, 111, 111t

funding costs and, 179ring-fencing and, 102for sovereign risk, 47

Evans, Charles L., 64EVT. See Extreme value theoryExchange rates

insurance sector and, 472nominal bilateral, 68, 75solvency risk and, 5

sovereign risk and, 183, 186See also Foreign currency risk

Expected cash flows, 421n13Expected default frequency (EDF), 157, 159

CDS and, 162–64sovereign risk and, 184

Expert Forum on Advanced Stress Testing Techniques, 12

Extended Fund Facility, of IMF, 347External terms of trade, 68Extreme value theory (EVT), 211

Fair value credit default swap (FVCDS), 157, 168Fair value model, for CDS, 197n27Federal Deposit Insurance Corporation, 189n6,

338Federal Reserve, 338Feedback effects

in FSAP liquidity stress test, 421for GDP growth, 50of liquidity risk, 189n7of macroeconomy, 50of solvency risk, 189n7of sovereign risk, 183, 202in stress test models, 389validity of, 189n7

Feedback loops, 4for bailouts, 93with liquidity and solvency risks, 24–25, 25bfor sovereign risk, 49, 49f

Finance Ministers and Central Bank Governors (FMCBG), 82

Financial market infrastructures (FMIs), 5stress tests for, 1, 22, 22n17, 24, 41–43

Financial openness, 317n6Financial repression, sovereign risk and, 225Financial sector assessment programs (FSAPs),

286n3, 328for CCPs, 5–6for credit risk shocks, 4DGI and, 83in France, 461FSSAs and, 18, 386–87GDP growth and, 32in Germany, 461for GFC, 1in GFSR, 12governance and, 419for insurance companies, 31liquidity stress tests and, 290in Luxembourg, 5, 26in Peru, 3for risk management, 4solvency risk in, 5sovereign risk and, 28, 184, 189in Spain, 4, 331fin Switzerland, 461in United Kingdom, 5in United States, 5, 26, 461See also Liquidity stress tests, FSAP; Solvency

stress tests, FSAPFinancial soundness indicators (FSIs), 85–86Financial stability analysis, 3, 8

liquidity stress tests in, 411–12network models in, 23b, 23f

Financial Stability Board (FSB)DGI and, 82, 83–84Early Warning Exercise and, 373


Index 519

Financial stability policy framework, 59–60

Financial System Stability Assessments (FSSAs)

FSAPs and, 18, 386–87FSAP liquidity stress tests and, 413FSAP solvency stress tests and, 375, 392–93,

399tSTeM for, 375, 393, 399t, 413

Finland, 288n14FIRB. See Foundation internal-ratings basedFiscal balance ratio, GFM and, 69Fiscal expenditure, GFM and, 70Fiscal policies, countercyclical

capital buffers and, 146, 255sovereign risk and, 183n1

Fiscal revenue, GFM and, 70Fisher-Tippert-Gnedenko theorem, 211Flight-to-quality, 287FMCBG. See Finance Ministers and Central

Bank GovernorsFMIs. See Financial market infrastructuresForeign currency risk, 90

in insurance sector, 477Foundation internal-ratings based (FIRB),

RWAs, 245, 245f, 253Fragility

detection of, 271–72exacerbation theorem, 272n15of GDP growth, 274nonlinearity and, 270n4shocks and, 271stress tests for, 269–83, 273f, 275t–77t, 278ftransfer theorem, 272n15See also Antifragility

Fragmentation, 312France

FSAPs in, 461FSAP solvency stress tests in, 375, 383–84GIIPS and, 289fHtM in, 189insurance FSAPs in, 485tinsurance sector of, 477RWAs in, 247, 385shocks in, 313Société Générale of, 271, 271n9

FSAPs. See Financial sector assessment programs

FSB. See Financial Stability BoardFSIs. See Financial soundness indicatorsFSSAs. See Financial System Stability

AssessmentsFull ring-fencing, 105, 109tFunding cost

capital requirements and, 168FSAP solvency stress tests and, 384solvency and, 421n10solvency risk and, 155–79, 161t–63t, 164f,

165f, 167t, 169t–78tsovereign risk and, 185, 186, 188spillover effects and, 49fstress tests for, 50, 281

Funding costs, risk aversion and, 177Funding liquidity risk, 416, 429Funding risk

of cross-border banking, 308in insurance sector, 472–73

Fuster, Andreas, 189n7FVCDS. See Fair value credit default swap

G20Antalya summit of, 83n4FSAP liquidity stress tests for, 412t, 413FSAP solvency stress tests for, 375t, 392ton tail risks, 270n3See also Data Gap Initiative

Galí, Jordi, 64GARCH. See Generalized autoregressive

conditional heteroskedasticityGDDS. See General Data Dissemination

SystemGDP. See Gross domestic productGDP growth, 274n18, 312

asset correlations and, 138–40, 152fcredit growth and, 138, 274, 274n18credit losses and, 136–37, 274, 274n18cross-country banking and, 5, 307–8, 316default rate and, 137, 137fin EMEs, 124feedback effects for, 50fragility of, 274FSAPs and, 32in GIIPS, 287preimpairment income and, 138public debt and, 283real GDP growth and, 135n44sensitivity to, 134, 134n42, 136t, 137, 137f,

138tshocks and, 277solvency tests and, 120spillover effects and, 5tail risks to, 3trading income and, 274See also Real GDP growth

General Data Dissemination System (GDDS), of IMF, 82, 91

General equilibrium, 202DSGE, 63, 189GFSR and, 26n22stress tests for, 3, 4, 56–57

Generalized autoregressive conditional heteroskedasticity (GARCH), 286–87

DCC and, 288–89, 297for GIIPS, 288–89, 289f, 292–93

Generalized extreme value (GEV), 185for CDS, 196–97, 202credit risk premium and, 211, 212, 213LRS for, 217

Gerali, Andrea, 64Germany

CDS in, 197FSAPs in, 461FSAP solvency stress tests in, 375insurance FSAPs in, 485t, 486tinsurance sector of, 477RWAs in, 247, 385safe haven in, 186n1shocks in, 313

GEV. See Generalized extreme valueGFC. See Global financial crisisGFF. See Global Flow of FundsGFM. See Global macro-financial modelGFSM. See Government Finance Statistics

ManualGFSR. See Global Financial Stability ReportGiannetti, M., 308Giesecke, Kai, 124GIIPS. See European peripheral countries

G-IIs. See Global systemically important insurance companies

Global financial crisis (GFC)crisis stress tests and, 340cross-border banking and, 307DCC and, 288n13DGI and, 82–93FSAPs for, 1FSAP solvency stress tests and, 384lessons from, 18–21liquidity risk in, 3, 155, 411right data for, 3SCAP and, 373solvency risk in, 5sovereign debt and, 383stress test models and, 18n13stress tests and, 11–12, 377

Global Financial Stability Report (GFSR)DGI and, 82–83, 83fFSAPs in, 12FSIs in, 85general equilibrium and, 26n22of IMF, 374

Global Flow of Funds (GFF), 84Globally Systemically Important Banks

(G-SIBs), 83, 87ring-fencing and, 103, 103n8

Globally Systemically Important Financial Institutions (G-SIFIs), DGI for, 87

Global macro-financial model (GFM)capital ratios and, 67data description for, 75data transformations, 70–71duration risk premium and, 66–73endogenous variables for, 64–70estimation, 70–71exogenous variables for, 70parameter estimation for, 71, 77t–78tresults of, 73, 74fscenario analysis for, 71–73, 72t, 73ffor stress test simulation, 63–75

Global risk aversion (VIX), 160LIBOR and, 160n18spillover effects and, 286, 288

Global systemically important insurance companies (G-IIs), 454, 454n5

Goldberg, Linda S., 102, 308Goldstein, Itay, 350Goodhart, Charles, 103n9Governance, 164

of crisis stress tests, 323in EMDEs, 225FSAPs and, 419RCAP and, 238n2risk, 92n20for stress tests, 3, 3n2

Government Finance Statistics Manual (GFSM), 96

Gray, Dale, 157, 286Greece

bailouts by, 329f, 356thaircuts in, 200n37shocks in, 314stress tests for, 31See also European peripheral countries

Greenlaw, David, 12Gross domestic product (GDP)

credit growth and, 274n18credit losses and, 274n18


Index520

Gross domestic product (GDP) (cont.)deficit-to-GDP ratio, 287solvency stress tests and, 292spillover effects and, 293stress tests and, 273surprises, cross-country banking and,

316–17, 316n3, 316ttail risks and, 272See also GDP growth; Real GDP growth

Group-structure approach, to ring-fencing, 103–4, 104f

G-SIBs. See Globally Systemically Important Banks

G-SIFIs. See Globally Systemically Important Financial Institutions

Haircutsfor credit risk premium, 214, 214n63in European sovereign debt crisis, 189in Greece, 200n37liquidity buffers and, 421liquidity stress tests and, 202n39for market risk, 206n44for sovereign risk, 189, 196–200, 201t–2t,

205–14, 205f, 209b–10b, 215f, 216f, 222t–23t, 383

in trading books, 421n11Haldane, Andy, 249n9Handbook on Securities Statistics, 88Hardy, Daniel, 3, 5, 27b, 292, 293Hasan, Iftekhar, 157, 421n10Hasan, Maher, 273, 286n4, 291n16Heath, Robert, 3Hedging, 155n2

BU for, 17tcounterparty risk and, 19n15with derivatives, 160RWAs and, 255spillover effects of, 418

Held-for-trading (HfT), 13sovereign risk and, 187–88, 196

Held-to-maturity (HtM), 13–14, 28, 29, 31credit risk and, 219in France, 189sovereign risk and, 187–88, 196in Spain, 189valuation gap of, 196n18

Hesse, Heiko, 4, 435HfT. See Held-for-tradingHigh-quality liquid assets (HQLAs),

415, 431Hirtle, Beverly, 323Hodrick, Robert J., 71Holding companies, 189n6Hong Kong Monetary Authority, 25bHousing loans, benchmarks for, 281n26HQLAs. See High-quality liquid assetsHtM. See Held-to-maturityHui, Cho-Hoi, 189n7, 189n7, 435Hülsewig, Oliver, 64Hurdle rates, 14, 29, 30b, 30f

crisis stress tests and, 346bHyperinflation, 183, 184, 186

IAG. See InterAgency Group of Economic and Financial StatisticsIAIS. See International Association of

Insurance SupervisorsIBS. See International Banking Statistics

Icelandcredit losses in, 134n43stress tests in, 270n5

ICF. See Implied-cash-flowICPs. See Insurance Core PrinciplesIFRS. See International Financial Reporting

StandardsIIP. See International investment positionIMF. See International Monetary FundImplied-cash-flow (ICF), 416, 419, 437–38,

442fIndia

GIIPS and, 289, 290fshocks in, 314, 315n2sovereign risk in, 183

Inflation ratein GIIPS, 287hyperinflation, 183, 184, 186insurance sector and, 470output, 64sovereign risk and, 183

Insurable interest, IAIS on, 459n19Insurance Core Principles (ICPs), of IAIS, 456,

460Insurance sector

asset-liability matching in, 459balance sheet in, 457f, 467n35, 480banks and, 456–59BU for, 462cash flow of, 459catastrophe risk in, 474concentration risk in, 472contagion risk in, 473–74CRAs of, 457credit risk in, 457, 472, 476–77demographic risk in, 474discount rate and, 469ECB and, 462n30equity risk in, 472, 487t–515tEuropean Systemic Risk Board and, 456n8exchange rates and, 472foreign currency risk in, 477FSAP solvency stress tests of, 453–516, 455f,

460f, 470f, 478b, 479f, 480f, 481t, 485t, 486t

funding risk in, 472–73funds to fulfill obligations of, 456n9G-SIIs, 454, 454n5inflation rate and, 470interest rates and, 472, 473b, 476liquidity risk in, 472–73macroprudential surveillance of, 454–56market-consistent valuation for, 467–69, 468fmarket risk of, 457off-balance-sheet and, 467n34reserve adequacy in, 474risk factors of, 472–77shocks and, 476single-period stresses of, 471sovereign debt and, 470STeM for, 463t–66tstress tests for, 22, 24, 41systemic risk of, 458bTD for, 462, 462n31underwriting risk in, 474, 474n53, 477underwriting shock and, 470

InterAgency Group of Economic and Financial Statistics (IAG), 82, 82n1

DGI and, 96

Interest ratescapital buffers and, 28elasticity of, 206n47GSM and, 66, 67, 69, 75insurance sector and, 472, 473b, 476public debt and, 195n17risk-free, 189, 189n11, 197, 203, 206–7robustness and, 288n12shocks, 195, 195n13, 213, 214n62solvency risk and, 5sovereign risk and, 187, 189on US Treasury bonds, 289

Internal ratings-based (IRB), 14n7asset correlations and, 138–40, 139b, 139fRWAs and, 121–22, 121n8, 140, 142t, 238,

239, 242, 249, 251n11, 255, 259, 261, 261f, 261n16, 387

solvency tests, 121sovereign risk and, 188, 195See also Advanced internal-ratings-based;

Foundation internal-ratings basedInternational Association of Insurance

Supervisors (IAIS), 454n5ICPs of, 456, 460on insurable interest, 459n19

International Banking Statistics (IBS), 82, 86, 90, 96

International Financial Reporting Standards (IFRS), 246–47, 246f

International investment position (IIP), 90, 96International Monetary Fund (IMF)

Action Plan for Strengthening Surveillance by, 88

Balance of Payments Manual of, 81Balance Sheet Analysis in Fund Surveillance

of, 92CDIS of, 90, 96CPIS of, 75, 82, 90, 96cross-country financial linkages and, 90debt sustainability analysis by, 185Early Warning Exercise of, 373Extended Fund Facility of, 347GDDS of, 82, 91GFF and, 84GFSM of, 89GFSR of, 374Handbook on Securities Statistics of, 88in IAG, 82n1Managing Director’s Action Plan for

Strengthening Surveillance of, 84Public Sector Debt Database and, 97Review of the FSAP by, 412SDDS of, 82SNA of, 81, 88–89Spillover Report of, 293TSR of, 84See also Financial sector assessment programs

International Organization of Securities Commissions (IOSCO), 22n17

International Swaps and Derivative Associa-tion, 197n27

IOSCO. See International Organization of Securities Commissions

IRB. See Internal-ratings-basedIreland

bailouts by, 328, 329fcrisis stress tests in, 323–24, 324t, 325t, 355,

356t–63tFSAPs for, 5


Index 521

PCAR in, 328, 330f, 339, 340, 344, 347, 353

shocks in, 313stress tests for, 31See also European peripheral countries

Israel, insurance FSAPs in, 485tItaly

shocks in, 313See also European peripheral countries

JapanFSAP solvency stress tests in, 375, 383–84,

471insurance FSAPs in, 469, 485t, 486tinsurance sector of, 477RWAs in, 385safe haven in, 186n1shocks in, 312, 314

Jensen’s inequality, 270, 272n15Jobst, Andreas A., 3, 5, 27b, 286, 411, 435, 454J.P. Morgan

Emerging Markets Bond Index of, 287, 288RiskMetrics of, 56

Kahn, 308Kalemli-Ozcan, Sebnem, 308Kaufman, George G., 103n9Kitamura, Tomiyuki, 156n4Kiviet, Jan F., 165Kiyotaki, Nobuhiro, 64Korea, GIIPS and, 289, 290fKovner, Anna, 157Krimminger, Michael H., 103n9

Labor force, GFM and, 68Laeven, L., 308Lagrange multiplier test (LMF), 165Langley, Paul, 323, 344Latin America, stress test models for, 18n14Lay, Kenneth, 321LCR. See Liquidity coverage ratioLehman Brothers, 285, 297, 313Le Leslé, Vanessa, 3Lender of last resort, liquidity and, 294n26Lending rate markup, GFM and, 70Leverage ratio, 140

RWAs and, 241, 241f, 242fLGDs. See Loss given defaultsLi, David, 88LIBOR. See London Interbank Offered RateLIDCs. See Low-income developing

countriesLinear combination of ratios of spacings (LRS),

for GEV, 217Linearity. See Concavity (linearity) effectsLiquidity

benchmarks for, 416, 423benchmark stress scenarios for, 299tCDS and, 200n38illustrative example for, 303indicators of, 416tlender of last resort and, 294n26spillover effects and, 287

ring-fencing and, 4stress tests for, 290–92, 291b, 292f, 293–94,

294f, 353Liquidity buffer

FSAP liquidity stress tests and, 421sovereign risk and, 186, 202

Liquidity coverage ratio (LCR)of Basel III, 416, 419, 422, 431–32, 433tFSAP liquidity stress tests and, 422

Liquidity metric monitor, 420Liquidity ratios

of Belgium, 420conceptualization of, 415fstress tests for, 278

Liquidity reporting profile, 419–20Liquidity risk, 14, 429

in economic cycle, 456n12feedback effects of, 189n7FSAP liquidity stress tests and, 413–16in GFC, 3, 155, 411in insurance sector, 472–73solvency risk and, 4, 24–25, 25b, 412, 435stress tests for, 270

Liquidity stress tests, 14–15cash-flow-based, 447in financial stability analysis, 411–12haircuts and, 202n39

Liquidity stress tests, FSAPasset encumbrance of, 420assumptions for, 439t–41tbanking book and, 422behavioral cash flows and, 421–22benchmarks for, 422as BU, 418calibration of, 420caveats to, 423–24communication of, 422–23concept of, 413–16contractual cash flows and, 421–22counterbalancing capacity and, 421counterparty risk and, 422coverage of, 418cumulative and noncumulative approaches

to, 415–16data for, 418–19feedback effects in, 421framework for, 416–23, 417tFSSAs and, 413FY2011-present, 425, 426t–28tfor G20, 412t, 413ICF and, 416, 419, 437–38, 442fliquidity buffer and, 421liquidity risk and, 413–14market liquidity risk and, 414–16methodology for, 422metrics for, 419–20NSFR and, 422premise for, 413publication of, 423reporting templates for, 443, 444t–46trisk horizon for, 420for S-29, 412–13, 412tscenario designs for, 419scope of, 418for SIB, 411–47solvency stress tests and, 420–21as TD, 421, 422in UK, 438

Liu, Liuling, 157, 421n10LLPs. See Loan loss provisionsLLRs. See Loan loss reservesLMF. See Lagrange multiplier testLoan loss provisions (LLPs), 159

sovereign risk and, 195Loan loss reserves (LLRs), 196

London Interbank Offered Rate (LIBOR), 160, 168

VIX and, 160n18Long-term nominal market interest rate, 66,

69, 75Long-Term Refinancing Operations (LTRO),

of ECB, 294, 328, 329f, 353Loss given defaults (LGDs), 125b, 128–29,

128f, 129t, 189n4CDS and, 209covered bonds and, 254downturn condition for, 195, 195n12FSAP solvency stress tests and, 384HtM and, 219real GDP growth and, 137–38risk-free interest rates and, 207RWAs and, 247, 248, 387solvency tests and, 122sovereign risk and, 188, 188n4, 195World Bank on, 129n30

Low-income developing countries (LIDCs)credit growth in, 130credit loss rates in, 126, 126n25, 127fLGDs in, 129preimpairment income in, 129–30retained earnings in, 138ROC in, 123satellite models for solvency tests and, 135solvency tests for, 121, 121n6

LRS. See Linear combination of ratios of spacings

LTRO. See Long-Term Refinancing OperationsLuxembourg

FSAPs for, 5, 26insurance FSAPs in, 485t

Macroeconomyfeedback effects of, 50stress tests of, 281

Macro-financial models, for FSAPs solvency stress tests, 392

Macroprudential solvency stress tests, 460bbenchmarks in, 187–89calibration of, 189–96, 190t–94tcapital impact of, 196empirical application of, 196–200, 198t–99tscope of, 185–87sovereign risk in, 183–225

Macroprudential surveillance, 3, 5of insurance sector, 454–56stress tests for, 11–50

Macroscenario stress tests, 14Malfeasance, 133, 133n38Managing Director’s Action Plan for

Strengthening Surveillance, of IMF, 84Market-consistent valuation

arbitrage and, 188CDS for, 186bfor insurance sector, 467–69, 468fprocyclicality and, 467for sovereign risk, 184–85, 188, 200

Market priceFSAP solvency stress tests and, 389–92, 391fof risk, 189n10stress tests and, 19b, 20t, 57, 58

Market riskcrisis stress tests for, 343thaircuts for, 206n44of insurance sector, 457


Index522

Market risk (cont.)of RWAs, 250–51, 252f, 387stress tests for, 270

Market-to-market (MtM), 14n4, 31, 188as benchmark, 29sovereign risk and, 225

Market valuationCAR and, 189sovereign risk and, 184

MAS. See Monetary Authority of SingaporeMayer, Eric, 64McElroy R2, 166, 166n26, 167McGuire, Patrick, 308Merton, Robert, 203n40Mexico

economic crisis of, 82GIIPS and, 289, 290finsurance FSAPs in, 485tinsurance sector of, 477sovereign risk in, 183

Microprudential/supervisory stress tests, 15, 56, 59

macroprudential stress tests and, 184Minoiu, Camelia, 308Monetary Authority of Singapore (MAS), 475bMonetary policy, quantitative easing in, 73Monte Carlo simulations, for SAD, 58n2Moody’s reports, 122, 128, 159n36

call option and, 159n16credit loss rates in, 135–36on EDF, 159on GDP sensitivities, 137high-yield spreads in, 287on LGDs, 137on RWA PDs, 247

Moore, John, 64Mora, Nada, 157Morgan, Donald P., 348MtM. See Market-to-marketMunoz, Sonia, 308Muto, Ichiro, 156n4

Neri, Stefano, 64Net foreign asset ratio, GFM and, 69Net government asset ratio, GFM and, 69The Netherlands

FSAP solvency stress tests in, 384insurance FSAPs in, 485tinsurance sector of, 477RWAs in, 385

Net stable funding ratio (NSFR)of Basel III, 416, 419, 422, 432FSAP liquidity stress tests and, 422

Network modelsin financial stability analysis, 23b, 23fin stress tests, 23b, 23f

New Keynesian DSGE. See Dynamic stochastic-general equilibrium

Nocera, Giacomo, 251n11Nominal bank lending interest rate, 67Nominal bilateral exchange rate, 68, 75Nominal ex ante portfolio return, 66Nominal policy interest rate, 66Nonfinancial corporate debt, 67Nonlinearity. See Convexity (nonlinearity) effectsNonperforming loans (NPLs)

credit risk and, 195n13crisis stress tests and, 349–50solvency tests and, 122, 123–24, 124t

sovereign risk and, 195stress tests for, 278

Nontraditional, noninsurance (NTNI), 22, 22n18

No ring-fencing, 105NPLs. See Nonperforming loansNSFR. See Net stable funding ratioNTNI. See Nontraditional, noninsurance

OECD. See Organisation for Economic Co-operation and DevelopmentOff-balance-sheet, 26

FSAP solvency stress tests and, 384IFRS and, 247insurance sector and, 467n34reputational risk and, 39RWAs and, 239

Office of Financial Research (OFR), 12Office of the Comptroller of the Currency,

U.S., 338Office of the Superintendent of Financial

Institutions (OSFI), 475bOFR. See Office of Financial ResearchOIS. See Overnight index swapOng, Li Lian, 3, 28n23, 411, 454Operational risk, 381

of insurance sector, 457, 465bRWAs and, 238, 239, 248

Organisation for Economic Co-operation and Development (OECD)

Basel I and, 249n10DGI and, 89, 96in IAG, 82n1Public Sector Debt Database and, 97RWAs and, 259

OSFI. See Office of the Superintendent of Financial Institutions

OTC. See Over-the-counterOura, Hiroko, 1–3, 5Out-of-the-money, 15n8Output price inflation, 64Outright Monetary Transactions, 353Overnight index swap (OIS), 160, 160n18, 168Over-the-counter (OTC)

CDS, 186n1derivatives, 84, 88

Pagano, Marco, 225Papaioannou, Elias, 308Parameter estimation, for GFM, 71, 77t–78tPartial ring-fencing, 105, 109tPCAR. See Prudential Capital Assessment

ReviewPDs. See Probabilities of defaultPedersen, Lasse Heje, 26Pension funds, 5

stress tests for, 24Peristiani, Stavros, 348Perri, Fabrizio, 308PGI. See Principal Global IndicatorsPoint-in-time (PIT), 14n6, 29Poland, 102, 109

GIIPS and, 289, 290fPolicy feedback, 25–26Popov, A., 308Portugal

insurance FSAPs in, 469, 485tinsurance sector of, 477See also European peripheral countries

Potential cash flows, 421n13PRA. See Prudential Regulatory AuthorityPreimpairment income, 27, 27b

in banking crises, 129–30, 130fcomponents of, 151tGDP growth and, 138satellite models for, 274

Prescott, Edward C., 71Price of commodities, GFM and, 69–70Price of equity, GFM and, 66–67, 75Primary fiscal balance ratio, 69Principal Global Indicators (PGI), 91–92Principles for Sound Stress Testing Practices and

Supervision, of BCBS, 39Probabilities of default (PDs), 125b

asset correlations and, 138CDS and, 209in downturn condition, 242FSAP solvency stress tests and, 384HtM and, 219risk-free interest rates and, 207RWAs and, 242–43, 247, 387solvency tests and, 122sovereign risk and, 188, 195S&P on, 247TTC, 188n4, 189n4

Procyclicalitycapital ratios and, 184, 225IRB and, 139bmarket-consistent valuation and, 467of public debt, 72rules of thumb and, 143–45, 144b, 145fRWAs and, 240tTTC and, 124n20

Profitfrom retained earnings, 138sovereign risk and, 225

Property marketsDGI and, 91, 97spillover effects in, 4in UK, 72

Prudential Capital Assessment Review (PCAR), in Ireland, 328, 330f, 339, 340, 344, 347, 353

Prudential Regulatory Authority (PRA), in UK, 462

Public debtin GIIPS, 287interest rates and, 195n17procyclicality of, 72sovereign risk and, 186, 225stress tests for, 274, 277, 278, 283See also Sovereign debt

Public Sector Debt Database, 97Puhr, Claus, 273, 286n4, 291n16

Quantitative easingby European Central Bank, 189in monetary policy, 73

Quasi-risk-weighted assets, 50Quasi-sovereign entities, sovereign risk

and, 183

RCAP. See Regulatory Consistency Assessment ProgramRCDS. See Running CDSReal ex ante corporate loan rate, 67Real GDP growth

benchmarks for, 139b


Index 523

country-specific interest rate shock and, 214n62

GDP growth and, 135n44interest rate premium and, 214LGDs and, 137–38satellite models for solvency tests and,

133–35SCAP and, 340

Real net profit, 67Real wage, 68Recessions

downturn conditions and, 195n12insurance sector and, 470, 471b

Regulatory bank capital ratio, 69Regulatory Consistency Assessment Program

(RCAP), 237–38, 238f, 238n2Reinsurance, 459n20Relative shadow price of capital, 66Reputational risk, 333, 484

off-balance-sheet and, 39Research Task Force (RTF), 12, 414tReserve adequacy, in insurance sector, 458b,

461, 470, 474Resti, Andrea, 251n11Retained earnings

profit from, 138stress tests for, 281

Return on assets, 278Return on capital (ROC)

credit losses and, 140–42impairment income and, 129solvency tests and, 122–23

Return on equity, 278Reverse stress tests, 460b, 460n1Review of the FSAP, by IMF, 412Richardson, Matthew, 157Ring-fencing

in Austria, 102n6Basel II/III and, 102consolidated stress tests and, 101–13in EU, 105–10, 106f, 107t–9t, 109tgroup-structure approach to, 103–4, 104fliquidity, spillover effects and, 4

Risk Assessment Model for Systemic Institu-tions, of Bank of England, 291b

Risk aversionCDS and, 158cost of funding and, 160EDF and, 169funding costs and, 177in GIIPs, 287–88in insurance sector, 458bmarket price of risk and, 189n10of sovereign risk, 189spillover effects and, 287VIX and, 160See also Global risk aversion

Risk-Based Capital, 467Risk-free interest rates, 189, 189n11, 197, 203,

206–7Risk governance, 92n20Risk horizon, 3, 27, 27b, 340, 371, 458b

for FSAP liquidity stress tests, 417t, 419–20for stress tests, 381

Risk interdependence, 32–34, 32b–33bRisk management

DGI for, 85–86, 85fFSAPs for, 4stress tests for, 1, 460b

RiskMetrics, of J.P. Morgan, 56Risk-neutral default probabilities, 30n27Risk-weighted assets (RWAs), 28n25, 187,

237–65accounting framework for, 246–47, 246fadvanced internal-ratings-based (AIRB),

243–45, 245f, 248f, 250, 253, 257tAIRB, 253antifragility of, 274asset correlations and, 138bank-related parameters for, 246f–51f,

247–51, 248n8, 249tBasel I/II and, 242Basel I/III and, 243–44Basel II and, 344–45, 385Basel III and, 50, 387calculation difference drivers of, 243–53calculation practice concerns for, 240tcapital buffers and, 238–39capital ratios and, 237, 238–41, 255CDMs and, 86CDS and, 177covered bonds and, 251, 254b, 254tCRAs on, 255credit risk of, 248–49, 249f, 249t, 259CT1 and, 159, 241, 242fdensity of, 241, 242f, 244t, 248feconomic cycle and, 247FIRB, 245, 245f, 253FSAP solvency stress tests and, 381, 385–87in GFC, 3IRB, 121–22, 121n8, 140, 142t, 238, 239,

242, 249, 251n11, 255, 259, 261, 261f, 261n16, 387

leverage ratio and, 241, 241f, 242fLGDs and, 247, 248, 387market risk of, 250–51, 252fmethodology and sample distributions for,

263, 263f, 264tOECD and, 259off-balance-sheet and, 239PDs and, 242–43, 247, 387preimpairment income and, 27RCAP and, 237–38, 238f, 238n2reform of, 253–55, 256f, 256t–57tregulatory framework for, 238–39, 243–46,

245f, 259–61, 260t–61t, 261fsolvency risk and, 241solvency stress tests and, 292solvency test for, 14, 120sovereign risk and, 196, 251S&P on, 247, 265, 265f, 266fstress tests for, 281in UK, 251, 253bvariability of, 242–43

Robustnessantifragility and, 272n12checks, for solvency risk, 168–72of cross-border banking, 317–18, 318f, 319tinterest rates and, 288n12problem, for stress tests, 57–58, 58n2

ROC. See Return on capitalRollover risk, 160Roulet, Caroline, 157RTF. See Research Task ForceRules of thumb

procyclicality and, 143–45, 144b, 145ffor satellite models, 133–40for solvency tests, 119–46, 121f, 136t, 146t

Running CDS (RCDS), 209Russia

capital ratios in, 109n20FSAP solvency stress tests in, 384GIIPS and, 289, 290fRWAs in, 385

RWAs. See Risk-weighted assets

S-25, 375t, 392tS-29, 412–13, 412tSA. See Standardized approachSAD. See System assets in distressSafe havens, 186n1Salman, Ferhan, 435Sapra, Haresh, 350Satellite models

for FSAP solvency stress tests, 387–88, 388f, 389f

panel regressions for, 274rules of thumb for, 133–40for solvency tests, 133–40

Savage, Lawrie, 157Savino, Vanessa, 348SCAP. See Supervisory Capital Assessment

ProgramScenario analysis, for GFM, 71–73, 72t, 73fSchaeck, Klaus, 123, 124tScheicher, Martin, 138Schmieder, Christian, 3, 4, 5, 25b, 27b, 138,

273, 286n4, 290, 291b, 291n16, 292, 293, 411, 422, 435, 454

Schmitz, Stefan W., 4, 5, 157, 158, 168, 189n7, 421n10

Schoar, Antoinette, 157Schoenmaker, Dirk, 102, 103n9Scholes, Myron, 203n40Schuermann, Til, 323, 350, 353Schumacher, Liliana, 1–3, 25b, 291b, 435Scope

of crisis stress tests, 339–40of FSAP liquidity stress test, 418of FSAP solvency stress tests, 378–79of stress tests, 26–28, 27b, 378–79

Scuzzarella, Ryan, 308SDDS. See Special Data Dissemination

StandardSDDS Plus. See Special Data Dissemination

Standard PlusSecond IMF Statistical Forum, 84–85, 84n8Securities depositories, 5Sell-off events, sovereign risk and, 187Sensitivity

of bank credit, 135to GDP growth, 134, 134n42, 136t, 137,

137f, 138ttests, for shocks, 383

Sessa, Luca, 64Shadow banks

DGI for, 86–87FSAPs liquidity stress tests and, 418

Shadow price of bank capital, 67–68Shocks

in Belgium, 313calibration of, 189–96, 190t–94tconvexity (nonlinearity) effects with, 206n49credit risk, 4cross-border banking and, 312–15, 313f–16fcurrency risk premium, 66equity risk premium, 66, 72t, 73


Index524

Shocks (cont.)fragility and, 271GDP growth and, 277in India, 315n2insurance sector and, 476interest rates, 195, 195n13, 213, 214n62in Ireland, 313sensitivity tests for, 383with sovereign risk, 184, 185, 189–96,

190t–94tstress tests for, 3tail risks and, 270in Turkey, 315n2See also Global financial crisis

Short-term debtCT1 and, 168solvency risk and, 157

Short-term nominal market interest rate, 66Short-term nominal price, in GFM, 64SIB. See Systemically important banksSiegmann, Arjen, 102Sigmund, Michael, 4, 5, 157, 158, 168, 189n7,

421n10Signoretti, Federico M., 64Simonelli, Saverio, 225Singapore

FSAP solvency stress tests in, 471insurance FSAPs in, 469, 485t, 486tinsurance sector of, 477MAS of, 475b

Single-period stresses, of insurance sector, 471Single Supervisory Mechanism (SSM), of ECB,

350Sixth Quantitative Impact Study, of BCBS, 387Smets, Frank, 63–64SNA. See System of National AccountsSociété Générale, 271, 271n9Solé, Juan, 5, 308Solvency

crisis stress tests for, 321–23, 322fdefined, 478funding cost and, 421n10sovereign risk and, 186spillover effects and, 287stress tests for, 290–92, 291b, 291n16, 292f,

293, 293f, 301t–2t, 322f, 376f, 378tSolvency I, 467Solvency II, 467, 480n62Solvency risk, 3

agent-based models for, 4CDS and, 157, 158–59credit spreads and, 5exchange rates and, 5feedback effects of, 189n7funding cost and, 155–79, 161t–63t, 164f,

165f, 167t, 169t–78tinterest rates and, 5liquidity risk and, 4, 24–25, 25b, 412, 435robustness checks for, 168–72RWAs and, 241short-term debt and, 157stress tests for, 270

Solvency stress testsassumptions in, 404tFSAP liquidity stress test and, 420–21See also Macroprudential solvency stress tests

Solvency stress tests, FSAPaccounting-based models for, 389, 391fbalance sheet and, 384, 389, 391f

as BU, 379, 388, 392, 394t–98t, 454capital standards for, 385–87communication of, 392credit growth and, 384cross-border banking and, 384data for, 380–81dividends and, 384for EU, 383in France, 383–84FSSAs and, 375, 392–93, 399tfunding cost and, 384FY2010-FY13, 392t, 400for G20, 375t, 392tGFC and, 384of insurance sector, 453–516, 455f, 460f,

470f, 478b, 479f, 480f, 481tin Japan, 383–84LGDs and, 384macro-financial models for, 392market-price-based models for, 389–92, 391fin Netherlands, 384off-balance-sheet and, 384PDs and, 384publication of, 392–93risk factors of, 383–84risk horizon for, 381in Russia, 384RWAs and, 381, 385–87for S-25, 375t, 392tsatellite models for, 387–88, 388f, 389fscope of, 378–79for SIB, 373–93sovereign risk and, 383for Spain, 384spillover effects and, 383, 389stress scenarios for, 381–83, 382fstress test models for, 388–92, 388f, 389f,

390t, 391fin Sweden, 384tail risks and, 383as TD, 379, 388, 454trading books and, 383–84in UK, 383, 384

Solvency testsbalance sheets and, 13–14, 13fbank characteristics and, 140, 141f, 141tin banking crises, 123–30, 124t, 125b,

125f–28f, 126t, 129t, 130f–33f, 132bGFM and, 63methodology and sources for, 122–23,

123n16NPLs and, 122, 123–24, 124tROC and, 122–23rules of thumb for, 119–46, 121f, 136t,

146tsatellite models for, 133–40simulations for, 140–43

South AfricaFSAP solvency stress tests in, 471insurance FSAPs in, 485tinsurance sector of, 477

Sovereign debtCCA for, 203crisis stress tests and, 325in EU, 286GFC and, 383insurance sector and, 470spillover effects of, 286, 287See also European sovereign debt crisis

Sovereign risk, 28n24, 31, 184fin AEs, 183bailouts and, 225bank bailouts and, 185, 225Basel III and, 184, 185b, 200BCBS and, 188–89benchmarks for, 187–89calibration of, 185, 189–96, 190t–94tcapital impact of, 196capital ratios and, 225causes of, 225CCA for, 203–4CDS and, 185, 202commodity price cycle and, 187counterparty risk and, 186credit risk and, 189, 195credit risk premium and, 185, 208default rate and, 183defined, 185beconomies with higher levels of, 186–87economies with low levels of, 186EDF and, 184in EMDEs, 183, 185, 187, 195, 200, 225EU system-wide stress test for, 47exchange rates and, 186expected and unexpected losses from,

187–89, 187ffeedback effects for, 183, 202feedback loops for, 49, 49ffinancial repression and, 225FSAPs and, 28, 184, 189FSAP solvency stress tests and, 383funding cost and, 185, 186, 188haircuts for, 189, 196–200, 201t–2t, 205–14,

205f, 209b–10b, 215f, 216f, 222t–23t, 383

hyperinflation and, 186interest rates and, 187, 189liquidity buffer and, 186, 202in macroprudential solvency stress tests,

183–225market-consistent valuation for, 184–85,

188, 200MtM and, 225profit and, 225public debt and, 186, 225risk aversion of, 189RWAs and, 251scope of, 185–87sell-off events and, 187shocks with, 184, 185, 189–96, 190t–94tsolvency and, 186spillover effects of, 183in state-owned banks, 187transparency of, 184

S&P. See Standard & Poor’sSpain

BU in, 347–48CDS in, 197crisis stress tests in, 323–24, 324t, 325t, 340,

341–44, 347–48, 347b, 355, 356t–63tFSAPs for, 4, 331fFSAP solvency stress tests for, 375, 384HtM in, 189insurance FSAPs in, 469, 485tRWAs in, 385TD in, 347See also European peripheral countries

Spaltro, Marco, 102, 157


Index 525

Special Data Dissemination Standard (SDDS), of IMF, 82

Special Data Dissemination Standard Plus (SDDS Plus), 91

Special purpose vehicles (SPVs), liquidity tests and, 15

Spillover effects, 4–5analysis of, 292–94, 293f, 294fbenchmarks for, 287in downturn condition, 285FSAP solvency stress tests and, 383, 389GARCH for, 286–87of GIIPS, 287–89, 289f, 290fof hedging, 418panel approach to, 287–88, 295t, 296tof sovereign debt, 286, 287of sovereign risk, 183from sovereigns, 49, 49fin stress tests, 285–303

Spillover Report, of IMF, 293SPVs. See Special purpose vehiclesSRISK, 157, 157n9SSM. See Single Supervisory MechanismStandardized approach (SA), 188Standard & Poor’s (S&P), 338f

on RWAs, 265, 265f, 266fon RWAs PDs, 247

State-owned banks, sovereign risk in, 187STeM. See Stress Test MatrixStiroh, Kevin, 323Stress Test Matrix (STeM)

for FSSAs, 375, 393, 399t, 413for insurance sector, 463t–66tfor insurance sector FSAPs, 487t–514t

Stress test models, 18, 19bbalance sheets, 19b, 20t, 57for credit risk, 50feedback effects in, 389for FSAP solvency stress tests, 388–92, 388f,

389f, 390t, 391ffor funding cost, 50GFC and, 18n13for Latin America, 18n14market price, 19b, 20t, 57, 58

Stress testsof AEs, 273–74agent-based models for, 59aggregation problem for, 57–58anatomy of, 1–6, 2fin Belgium, 478bbenchmarks for, 120best practices for, 1–3, 21–36, 37tcalibration of, 31–34, 32b–33bfor central banks, 3closed-form expression in, 272n14communication in, 34–35for contagion risk, 270–71context for, 375–77coverage for, 21–24for credit growth, 281for credit losses, 281data for, 380–81defined, 12–18DGI and, 92–93in financial stability policy framework, 59–60for FMIIs, 1, 22, 22n17, 24, 41–43for fragility, 269–83, 273f, 275t–77t, 278ffor funding cost, 281GDP and, 273

for general equilibrium, 56–57GFC and, 11–12, 377governance for, 3, 3n2in Iceland, 270n5implementation of, 3n2, 5for insurance sector, 22, 24, 41interpretation of, 28–31limitations of, 35–36for liquidity, 290–92, 291b, 292f, 293–94,

294f, 353for liquidity risk, 270of macroeconomy, 281for market risk, 270motivation for, 1network models in, 23b, 23fnew generation of, 56–58operational implications of, 36–37outcomes of, 3n2for pretax income, 281principles and practices for, 11–50for public debt, 274, 277, 278, 283for retained earnings, 281right data for, 3risk horizon for, 381risk interdependence in, 32–34, 32b–33bfor risk management, 1, 460brobustness problem for, 57–58, 58n2for RWAs, 281scope of, 26–28, 27b, 378–79sensitivity of, 297n8for shocks, 3simulation of, GFM for, 63–75for solvency, 290–92, 291b, 291n16, 292f,

293, 293f, 301t–2t, 376f, 378tfor solvency risk, 270spillover effects in, 285–303stylized design of, 286, 286fsupervisory guidelines for, 39for systemic risk, 55–60tail risks in, 31–34, 32b–33b, 269–83transmission of, 24–26, 24ftypology of, 15–18, 16tin United States, 111, 111tWEO and, 273See also Financial Sector Assessment

Program; Supervisory Capital Assessment Program; specific types

Strike price, 15n8Supervisory Capital Assessment Program

(SCAP), 12, 15, 34, 45, 46, 46n50, 46n52, 272n13, 321, 338

AQR in, 339, 350BU in, 339disclosure of technical details of, 348GFC and, 373standardization of assumptions for, 344success of, 325–28, 327fTD in, 339transparency of, 345–47See also Comprehensive Capital Analysis and

ReviewSurveillance agenda, for DGI, 84, 88–89Surveillance stress tests, 389Sweden

FSAPs for, 5FSAP solvency stress tests in, 375, 384

SwitzerlandFSAPs in, 461GIIPS and, 289f

insurance FSAPs in, 469, 485t, 486tinsurance sector of, 477RWAs in, 247shocks in, 313

System assets in distress (SAD), 58n2Systemically important banks (SIB), 239

BCBS on, 380crisis stress tests for, 339in FSAP liquidity stress tests for, 411–47solvency stress tests in FSAP solvency stress

tests for, 373–93Systemic Contingent Claims Analysis, 392Systemic risk

of insurance sector, 458bstress tests for, 55–60transparency of, 184

System of National Accounts (SNA), 81, 88–89

Tail risksbenchmarks for, 21concavity (linearity) effects of, 270, 271f,

271n8, 278n23convexity (nonlinearity) effects of, 270,

271–72, 274, 274n17FSAP solvency stress tests and, 383G20 on, 270n3to GDP growth, 3shocks and, 270stress tests for, 31–34, 32b–33b, 269–83

Takei, Ikuo, 156n4Taleb, Nassim N., 270, 270n4, 272n12,

272n15, 297n8Tangible common equity, 239–41, 241fTangible total assets, 239–41, 241fTarazi, Amine, 157TCE. See Tangible common equityTD. See Top-down stress testsThrough-the-cycle (TTC), 14n6, 29, 188n4,

189n4procyclicality and, 124n20

Tier 1 capitalcomparison table for, 405, 406t–7tFSAP solvency stress tests and, 385SCAP and, 325, 328solvency stress tests and, 292spillover effects and, 294

Too-big-to-fail, 83Top-down stress tests (TD), 17t, 338, 340

crisis-management stress tests and, 46FSAP liquidity stress tests as, 418, 421,

422FSAP solvency stress tests as, 379, 388, 454for insurance sector, 462, 462n31in Spain, 347

Traclet, Virginie, 435Trade balance ratio, 69Trade openness, 287Trading books

arbitrage for, 259capital requirements for, 250FSAP solvency stress tests and, 383–84haircuts in, 421n11sovereign risk and, 188

Trading income, 132b, 132fGDP growth and, 274

Trading losses, credit losses and, 132bTransmission

channels, sovereign risk and, 183, 184of stress tests, 24–26, 24f


Index526

Transparency, 12, 31, 84, 133, 374with AfS, 196n19with cross-border banking, 319GDDS and, 91GFC and, 3MtM and, 188of sovereign risk, 184of systemic risk, 184

Triennial Surveillance Review (TSR), of IMF, 84Tsatsaronis, Kostas, 340TSR. See Triennial Surveillance ReviewTTC. See Through-the-cycleTucker, Paul, 88Turkey, 109

crisis stress tests in, 323GIIPS and, 289, 290fshocks in, 314, 315n2

Udell, G. F., 308UFR. See Ultimate forward rateUK. See United KingdomUltimate forward rate (UFR), 467n35Unconsolidated balance sheets, 4Underwriting risk, 474, 474n53, 477Underwriting shock, 470United Kingdom (UK)

bank credit in, 73CDS in, 197FSAP liquidity stress test in, 438FSAPs for, 5FSAP solvency stress tests in, 375, 383, 384

GFM for, 72–73insurance FSAPs in, 486tPRA in, 462property markets in, 72RWAs in, 247, 251, 253b, 385shocks in, 313

United Statescapital ratios in, 109n20CCAR in, 272n13, 350, 373crisis stress tests in, 323–24, 324t, 325t,

351t–52t, 355, 356t–63tDFA in, 17n12, 24n20, 246, 251, 321,

350Dodd-Frank Act in, 17n12, 24n20FSAPs for, 5, 26, 461FSAP solvency stress tests in, 375GFM for, 73insurance FSAPs in, 485t, 486tinsurance sector of, 477PDs in, 247Risk-Based Capital in, 467RWAs in, 247–48safe haven in, 186n1SCAP in, 15, 34shocks in, 312, 314stress tests in, 111, 111tSee also Supervisory Capital Assessment

ProgramUS Treasury bonds

interest rates on, 289yield curve for, 288n12

Valderrama, Laura, 4, 5, 158, 168, 189n7, 421n10Value at risk (VaR), 15, 188Van den End, Jan Willem, 291b, 291n1, 435Van Lelyveld, Iman, 102, 308VaR. See Value at riskVitek, Francis, 63, 71, 274, 286n4VIX. See Global risk aversion

Wage rigidities, in GFM, 64Wald test, 317, 317n4Webber, Lewis, 58n2Wehrhahn, Rodolfo, 157Weighted average, GFM and, 65WEO. See World Economic OutlookWillison, Matthew, 58n2Wollmershäuser, Timo, 64Wong, Eric, 189n7, 435World Bank, 11

GFM and, 75in IAG, 82n1on LGDs, 129n30Public Sector Debt Database and, 97

World Development Indicators, GFM and, 75World Economic Outlook (WEO)

GFM and, 75on real GDP growth, 134on spillover effects, 292–93stress tests and, 273

Wouters, Raf, 63–64

Zhang, Gaiyan, 157, 421n10


Stress Testing - IMF eLibrary - International Monetary Fund

Documents

Transcript of Stress Testing - IMF eLibrary - International Monetary Fund